Fall 2008 Biomedical Informatics
Table of Contents
1 . The disciplines of biomedical informatics; biomedical informatics as a career
1.1 Summary of lecture on August 25th:
1.2 Summary of readings for August 25th
1.2.1 Hersh - Who are the Informaticians?
1.2.2 Shortliffe - Biomedical Informatics; Chapter 1
1.3 Summary of lecture on August 27th
1.4 Summary of readings for August 27th
1.4.1 Sagotsky - Life Sciences and the web
1.5 Helpful links
1.5.1 The NHS
1.5.2 Misc
1.5.3 More Info on HIT
1.5.4 More Info on the Semantic Web
1.6 Education and Career Possibilities
2 Week 2 Survival guide: things you should know in order to navigate the field
2.1 Acronyms Defined: Organizations/Associations/Journals
2.2 Acronyms Defined: Projects/Technologies
2.3 Approximate top 6 Medical Informatics Programs around the country
2.4 WORD document of Electronic Health Records (EHR) Terms
3 Week 3 The biomedical world: healthcare systems
3.1 Lecture Review
3.1.1 Healthcare System
3.1.2 Rising Costs
3.1.3 Healthcare Payers
3.1.4 Information Flow in Healthcare Systems
3.2 Articles
3.2.1 National Health Information Infrastructure
3.3 Additional resources:
4. Week 4 Biomedical data: what is/how it is communicated
4.1 Shortliffe Chapter 2
4.1.1 What are Medical Data?
4.1.1.1 What Are the Types of Medical Data?
4.1.1.2 Who Collects the Data?
4.1.2 Uses of Medical Data
4.1.2.1 Create the Basis for the Historical Record
4.1.2.2 Support Communication Among Providers
4.1.2.3 Anticipate Future Health Problems
4.1.2.4 Record Standard Prevention Measures
4.1.2.5 Identify Deviations from Expected Trends
4.1.2.6 Provide a Legal Record
4.1.2.7 Support Clinical Research
4.1.3 Weakness of the Traditional Medical Record System
4.1.3.1 Pragmatic and Logistical Issues
4.1.3.2 Redundancy and Inefficiency
4.1.3.3 Influence on Clinical Research
4.1.3.4 The Passive nature of Paper Records
4.1.4 The Structure of Medical Data
4.1.4.1 Coding Systems
4.1.4.2 The Data–to-Knowledge spectrum
4.1.5 Strategies of Medical Data Selection and Use
4.1.5.1 The Hypothetico-Deductive Approach
4.1.5.2 The Relationship Between Data and Hypotheses
4.1.5.3 Methods for Selecting Questions and Comparing Tests
4.1.6 The Computer and Collection of Medical Data
4.2 Kolodner Article- Health Information Technology: Strategic Initiatives, Real Progress
4.2.1 A multifaceted, Balanced Agenda
4.2.2 Recent Activities and Progress
4.2.2.1 Adoption
4.2.2.2 Collaborative Governance
4.2.2.3 Privacy and Security
4.2.2.4 Interoperability
4.2.3 The Interaction of Technology and Policy
4.2.4 Working Together to Accelerate Progress
4.3 Olivier Bodenreider Article: "The United Medical Language System (UMLS): integrating biomedical terminology"
4.3.1 UMLS: What is it? What are the main components of UMLS?
4.3.2 What included in UMLS?
4.3.3 How UMLS is organized?
4.3.4 What users can do with UMLS?
4.3.5 What tools does UMLS have?
4.4 de Keizer article: "Understanding Terminological Systems I: Terminology and Typology"
4.4.1 Introduction
4.4.2 Terminology and Definitions (objects, concepts and designations)
4.4.2.1 Relationships between Concepts
4.4.2.2 Definitions of Concepts
4.4.3 Typology of Terminological Systems
4.4.4 Typology of Existing Terminological Systems
4.4.4.1 ICD-9-CM and ICD-10
4.4.4.2 NHS clinical terms
4.4.4.3 SNOMED
4.4.4.4 UMLS
4.4.4.5 GALEN
5. Week 5 Biomedical data (metadata/ontologies) & HIS overview
5.1 Metadata/Ontologies
5.1.1 Summary of lecture on 9/22/08
5.1.2 Summary of Bodenreider article
5.1.3 Summary of Cimino article
5.2 HIS Overview
5.2.1 Elements and Components of HIS
5.2.2 Key functions of the HIS
5.2.3 Options and Factors for Building the HIS
5.2.4 Summary from Group Exercise
6. Week 6 HIS case studies (John Wadsworth: Intermountain; Xiufen Li: UUHSC)
6.1 Lecture about UUHSC
6.2 Summary of Murff article
6.3 Acronyms Defined
6.4 Notes from wiki for a priori
7. Week 7 HIS case study and evaluation 7.1 Introduction to the VA Electronic Medical Records System VistA and its GUI interface CPRS
7.2 Major distinguishing characteristics of the VA Health Care System?
7.3 Some definitions:
7.4 Distinguishing characteristics of the VA IT Infrastructure?
7.5 VistA/CPRS
7.6 VistA Technology Stack (see slide on this in the class presentation)
7.7 CPRS is highly customizable
7.8 Installation of CPRS Started with one VA site
7.9 CPOE in the VA
7.10 Q&A Session:
7.11 Demonstration of CPRS
7.12 The future of VistA and CPRS?
7.13 Additional links, further reading, and papers of interest:
9. Week 9 Overview of decision support systems and patient safety (Shuying Shen: imaging informatics; Anthony Garcia: CDSS overview; Bryan Gibson: CDSS example: patient safety)
9.1 Imaging Informatics
9.1.1 Common usage of computers in radiology
9.1.2 CT, MRI and nuclear medicine are the main digital imaging modalities.
9.1.3 Other usage of computers in radiology
9.1.4 The problems of data overload
9.1.5 Advantages of PACS
9.1.6 Radiologic Process
9.1.7 Imaging Modalities
9.2 Requirements for an optimal and functional clinical decision support system
9.3 Clinical decision support system (CDSS) definition
9.4 Why the necessity of clinical decision support systems?
9.5 Kawamoto’s study
9.6 Fifteen features of clinical decision support systems
9.7 The seven potential explanatory features of clinical decision support systems
9.8 Main identified types of clinical decision support systems
9.9 Main features required for an optimal and functional clinical decision support system
9.10 Additional features that could improve clinical decision support systems
9.11 Conclusion
10. Week 10 DSS special topic presentations (team): case studies 10.1 Use of Data Warehousing as a Decision Support Tool For Patients With Depression
10.2 Order Sets
10.2.1 Infobutton Implementation at Intermountain HealthCare
10.2.2 A. What is Infobutton?
10.2.3 B. What the basic goal of Infobutton and why is that goal important?
10.2.4 C. Is Infobutton used in a real clinical setting?
10.2.5 D. What are the pros and cons of Infobutton, and evaluation in the articles?
10.3 Infobutoon Standardization.
10.9 Arden Syntax
10.9.1 What is Arden Syntax
10.9.2 What is the basic goal of the Arden Syntax and why is that goal important?
10.9.3 Is Arden Syntax used in a real clinical setting?
10.9.4 What are the pros and cons of Arden Syntax, and the future of Arden Syntax?
10.9.5 E. Summarization (Description) of MLM and Arden Syntax
11. Week 11 Translational informatics and FURTHeR
11.1 Lecture Summary
11.2 Kuhn et al. Summary
11.3 Butte Summary
11.4 FURTHeR
13 Week 13
13.1: Natural language processing in biomedicine
13.1.1 Problem Set: Mining Clinical and Biomedical Text
13.1.1.1 Clinical Text
13.1.1.2 Biomedical Text
13.1.2 Answer to mining free text data: Natural Language Process.
13.1.3 Approaches to Natural Language Processing
13.2: Nursing Informatics
13.2.1 Lecture Review
13.2.2 What is nursing informatics?
13.2.3 The evolution of nursing informatics
13.2.4 Why is nursing informatics so important
13.2.5 Nurse Decision Making
13.2.6 Nursing Terminology
13.2.7 Nursing Minimum Data Set Systems
13.3 Summary for Ozbolt on a brief history of nursing informatics
13.4 Summary for Bakken et al on the future of nursing informatics
13.5 Additional Resources
14 Week 14
14.1: Informatics and Cognitive Science
14.1.1 Article Summary: "Human-Computer Interaction in Health Care Organizations" by Nancy Staggers
14.2.1 Health Context
14.3.1 Additional readings:
14.2: Informatics in Clinical Education
15.week
15.1
15.1.1 Institutional Review Board (IRB)
15.1.2 Health Insurance Portability and Accountability Act (HIPAA)
15.1.3 Challenges of De-identifying Free-text Clinical Notes
15.1.4 Why is de-identification necessary?
15.1.5 A reliable reference standard is the cornerstone of system evaluation
15.1.6 Three specific examples of Automated NLP Methods
15.2 Semantic Webs,Semantic Networks, and A Use Case(Cleveland Clinic's application)
15.2.1 Why do we need Semantic Web?
15.2.2 Common Semantic Networks:
15.2.3 Cleveland Clinic’s application of semantic web technology
15.3 Goals of the Semantic Web used at the Cleveland Clinic.
15.3.1 The key Semantic Web advantages at the Cleveland Clinic are:
15.3.2 SemanticDB Processing
15.3.3 Conclusions of the Semantic Web of Cleveland Clinic’s Adoption[47]
15.4 An Introduction to Handheld Computing in Medicine.
15.4.1 Some common characteristics of Handheld computing
15.4.2 Why Physicians Need Handheld Computing in Medicine
15.4.3 What is the Basic Goal of Handheld Computing
15.4.4 Why is That Goal Important?
15.4.5 Pattern of PDA Usage
15.4.6 What are handheld computers good for?
15.4.7 What are they not so good for?
15.4.8 Handheld computers currently have various limitations, including
15.4.9 Doctors' experience with handheld computers in clinical practice.
15.4.10 Additional readings:
16 Week 1616.1 Grid Computing, PACE and Decision Support16.1.1 Grid Computing
16.1.2 PACE System
16.1.3 Clinical Decision Support Models16.
16.1.4 Clinical Decision Support Systems
16.1.5 Links
16.2 Drug Decision Support System presented by Anusha Muthukutty and Sharanya Raghunath
16.2.1 What is Drug Decision Support?
16.2.2 Different Stages of Decision Support
16.2.3 Decision Support in Drug Allergy checking
16.2.4 Basic Dosing Guidelines
16.2.5 Formulary Decision Support
16.2.6 Duplicate Therapy Checking
16.2.7 Drug–Drug Interaction Checking
16.3 Advanced Clinical Decision Support
16.3.1 Advanced Dosing Guidance in CPOE
16.3.2 Advanced Guidance for Medication-associated Laboratory Testing
16.3.3 Advanced checking of drug-disease interactions and contraindications
16.4.4 Advanced drug pregnancy alerting
16.4 Case Study: The impact of academic detailing and a computerized decision support system (CDSS) on prescribing medication at the point of care
16.4.1 Background and Setting
16.4.2 Research question
16.4.3 General Plan of Attack
16.4.4 Methods
16.4.5 Computerized decision support
16.4.6 Data collection
16.4.7 Outcome Measures
16.4.8 Results
16.4.9 Discussion
16.4.10 Limitations
16.4.11 Conclusion
16.4.12 Additional readings:
16.5 Personal Health Records, Microsoft and Google presented
16.5.1 What is a Personal Health Record or PHR?
16.5.2 Why bother with an e-PHR?
16.5.3 What are the types/models of PHR?
16.5.4 Points of concern
16.5.5 Evaluating PHR
16.5.6 Tour of Microsoft
16.5.7 Tour of Google
16.5.8 Types of PHR
16.5.9 PHR Platforms
16.5.10 Afterthoughts
16.5.11 Future of PHR?
16.5.12 Additional Reading
16.6 Drug Coding Standards
16.6.1Additional Reading
Week 1 The disciplines of biomedical informatics; biomedical informatics as a career
Summary of readings for August 25th
Who are the Informaticians?
Growing Evidence supports the use of this technology in regards to clinical decision support to improve health care safety and quality. Common issues discussed are cost, interference with clinical workflow, and technical support. We don't have a full knowledge of the background of Hospital Information Technology (HIT) personnel; specifically those who have clinical backgrounds. American Medical Informatics Association (AMIA) wants at least one Dr and 1 RN (Registered Nurse) at every hospital to be trained in medical informatics to guide implementation of HIT. HIT workforce: Clinicians, Information Technology (IT) professionals, Healthcare Informatics Management (HIM) professionals, and health science librarians. Findings indicate that regardless of size, each IDS (integrated delivery system) had about 1 IT staff per 56 non-IT employees. HIM profession had undergone significant changes during switch from paper records to Electronic Health Records (EHRs). Skills need to transition from managing paper folders to managing electronic data. Medical Informatics education could prove to be a very useful pathway to train upcoming informaticians. There has been no formal analysis of informatics in the workforce, and few practicing informaticians have had any formal training. No official "framework" of informatics practice, with lines between job titles fuzzy. There have been attempts to define competencies in informatics for clinical practitioners in the US. The NHS in the United Kingdom (UK) has defined them. Is Informatics a profession? It has some characteristics of a profession; such as strong knowledge base, intellectual technique, a highly developed code of ethics, and a sense of community. It does not, however meet all the definitions. Author believes that national certification would help establish it more concretely. Medical Informatics and HIT in general is a heterogeneous field at the intersection of health care and IT, with very diverse people working in it. Although the growing role of HIT in improving the quality, safety, and cost of health care is undisputed, who to teach what and when is currently a heavily debated topic.
Shortliffe - Biomedical Informatics; Chapter 1
Informatics is a new discipline, the result of applying computers to Medicine and Biology. Technology (Computers especially), have helped the field of medicine make discoveries and improve exponentially. It is the hope that computers will help elminate the inefficiencies and frustrations associated with paper-based medical records and the inadequate access to clinical information clinicians encounter. EHRs/EMRs (Electronic Health Records/Electronic Medical Records) have already helped develop integrated clinical workstations. It creates single-entry points for computational tools to assist clinicians not only with clinical matters, but also with administrative and financial topics. There are many inadequacies of traditional paper records, especially in modern medicine. There are to many to touch on all of the problems, but they include communication between providers, difficulty obtaining information, decrease in productivity, the complexity and sensitive nature of medical records. These also cause problems in switching from paper to electronic records. CPRs (Computer-based patient records) also help support clinical trails. There are recurring issues that need to be addressed. (1) The need for standards in the area of clinical terminology (2) concerns regarding data privacy (3) challenges of data entry by physicians and (4) difficulties associated with the integration of record systems with other information resources in the health care setting. The enterprise-wide network is called the intranet and it can create problems with integrating patient records, depending on the network. the NSF (National Science Foundation) took over the task of running the principal high speed backbone network in the 1980s, and has since been working on extending the record collection and implementation beyond the single institution. Section 1.1.4 has the model of the integrated disease surveillance The de facto standard for communicating and sharing information is Health Level 7 (HL7). The National Health Information Infrastructure (NHII) is an attempt to link all practices and practitioners in the country. There are many different branches of informatics or, more broadly, medical computer sciences, including information science, cognitive science, information theory, biocomputation, medical informatics, medical computing,Artificial Intelligence (AI), and biomedical informatics. There are links below for all of these fields. In the late 1990s, the NIH (National Institutes of Health) recommended undertaking an initiative called the Biomedical Information Science and Technology Initiative (BISTI). The history of informatics goes all the way back with the history if the computer, back to World War II, and is closely tied to the history of epidemiology and public health and Computer Science. The nature of Medical Information is very complex and raises unique problems and concerns. Although computers are increasing prevalent in our everyday lives, there are still many issues arising in the integration of biomedical computing and medical practices. Hopefully we will more fully address these as the course continues!
Summary of lecture on August 27th
* The DBI "Three Circles" again with examples and faculty who focus on those fields. This class will focus on Clinical Informatics with some Genetic Epidemiology popping in from time to time.
* What are the three circles?
o Clinical Informatics
o Public Health Informatics
o Clinical Research and Translational Informatics
* What is clinical informatics?
o Focuses on computer applications that address medical data. It is a combo of CS, information science and clinical science.
o Can be divided into two categories: Theory and Application
What jobs to can informaticians do and in what fields? (answered by students in class) What jobs to grads of the department get? where are they hired? and what is their pay scale? Review and contribute to the Study Guide!
Summary of readings for August 27th
Sagotsky - Life Sciences and the web
There is no debate that over the past decade computers, the internet, and the World Wide Web have revolutionized the way people communicate and share information. The question is how useful these changes are. One of the biggest challenges is the quality of the accessible data. Wikipedia, for example, has an average of 3.86 factual errors per article, compared to Encylopedia Britannica (2.92). It is also harder to determine what published has been peer reviewed, and what has not. Although an author may post one's methods along with the findings, there is still the question of origin of the data.
The viral nature of digital information lets information go across the world almost instantly, but the context, legal rights, and dissemination of facts is virtually unchecked. There are ways to help prevent the problems, but many are ineffective and/or rarely used due to complexity vs. perceived benefit. There is also no standard format for the exchange of data over the web, which leads to the predicable problems.
The Semantic web: another major issue is that beyond the formatting and rendering instructions contained in byte streams, the remaining information is, for the most part, totally unintelligible to the computers facilitating them today. a way to rectify this would be to implement what the author calls the "Semantic web".
Key ingredients of the semantic web: Triple form graphs, Web Ontology Language (OWL), Service Oriented Architectures (SOA), Web 2.0, and the SPARQL standard. There are many justifications for implementation and just as many criticisms against it.
One major technology that has influenced the Web 2.0 revolution is the wiki. While, as mentioned before, they are not praised for their accuracy, they do excel in versatility. The usage can scale as needed. It also gets the average web surfer grows more accustomed to publishing data online rather than just reading it. Blogs are another major influence.
Their content may be more up-to-date than research papers. They are admitted less balanced and unedited. Biomedical arenas have already embraced the web, including
· The Alzheimer Research Forum
· The HIV Forum, and the Cancer Biomedical Informatics Grid.
These types of web facilitates collaborative biomedical research, education, and outreach unlike anything we have seen before. It encourages and enables global exchange and multidisciplinary interaction. Although the system is not perfect, the biomedical research community will continue to innovate and will find ways to deal with current and future problems.
Helpful links
The NHS
* NHS Choices
Misc
* HIMSS - Electronic Health Record (EHR)
* Medical Records Institute
* A Clinical Aspect of the Computer-Based Patient Record
* The Need for Technical Solutions for Maintaining the Privacy of EHR
* Integrated disease surveillance
* Integrated Disease Surveillance Project - NICD
* The National Health Information Infrastructure, NHII
* Telemedicine.com
* American Telemedicine Association
* Edinburgh University, Division of Informatics, Artificial Intelligence
* Biomedical Information Science and Technology Initiative (BISTI)
* DMICE: Department of Medical Informatics and Clinical Epidemiology
* Public Health Informatics Institute
* Welcome to Clinical Informatics
* Center for Clinical and Translational Informatics
* The Society for Imaging Informatics in Medicine
* What is Structural Informatics?
More Info on HIT
* Characterizing the Health Information Technology Workforce
* A Closer Look at Healthcare Workforce Needs in the West
* iHealthBeat - Reporting Technology's Impact on Health Care
* Bioinformatics Opportunities for Health Sciences Librarians and Information Professionals
* The Educational Role of Health Sciences Librarians
* Information Technology in the Future of Health Care
* The Future of Health Information Technology
* Future of Health IT: Trends and Scenarios: BBC
More Info on the Semantic Web
Resource Description Framework (RDF): Concepts and Abstract Syntax
* More info on OWL:
OWL Web Ontology Language Overview
* More info on SPARQL standard:
SPARQL Query Language for RDF
* More info on SOA:
Web Services and Service-Oriented Architectures
* More info on Web 2.0 in healthcare:
Web 2.0 Revolution Underway in Healthcare
* Web 2.0:
Web 2.0 Revolution: Power to the People
Education and Career Possiblities
* Phds.org
* AMIA
* Medhunters.com
* Biohealthmatics.com
* U.S. Department of Labor - Occupational Employment Statistics
* University of Utah Biomedical Informatics Department
* Medical Informatics Meets Medical Education - Shortliffe
Week 2 Survival guide: things you should know in order to navigate the field
Acronyms Defined: Organizations/Associations/Journals
* NIH = National Institutes of Health (http://www.nih.gov)
o Part of the Department of Health and Human Services
o A very large government agency ($30B/yr budget), comprised of 27 institutes and centers and has over 18,000 employees
o Most of the money is invested into research via grants
* NLM = National Library of Medicine (http://www.nlm.nih.gov)
o One of the institutes of the NIH.
o One of the primary sources of funding for BMI students/faculty
+ Example: The T15 training grants that support fellowships at about 20 different informatics programs around the country)
* AMIA = American Association of Medical Informatics (http://www.amia.org)
o Publishers of JAMIA (Journal of the American Association of Medical Informatics)
+ One of the best rated journals in the field and common journal for BMI students/faculty to publish in.
o Organizers of the annual symposium and spring congress, a very important and relevant conference in the BMI students/faculty.
* IMIA = International Medical Informatics Association (http://www.imia.org)
o Similar to AMIA, but on an international scope.
o Not as popular in the United States
o Publishers of IJMI (http://www.elsevier.com/wps/find/journaldescription.cws_home/506040/description#description)
* ITS = Information Technology Services (http://uuhsc.utah.edu/ITS/)
o The local IT department specifically for the UUHSC (often referred to as the "upper campus")
o Separate/independent from the IT department for the rest of the University (aka the "lower campus")
* IRB = Institutional Review Board
o A committee that must review and approve all patient centric research to ensure that it is done lawfully and ethically.
Acronyms Defined: Projects/Technologies
* FURTHeR = Federated Utah Research and Translational Health e-Repository
o A project of the CCTS to build an information architecture to support research and collaboration in a federated (ie non-centralized) way.
* caBIG = cancer bio-informatics Grid (https://cabig.nci.nih.gov/)
o A project of the national cancer institute (NCI) that links together many different databases and researchers.
o It is also federated (non-centralized)
* UPDB = Utah population database (http://www.hci.utah.edu/group/sharedFacilities~/ccsg/UPDB.jsp)
o A one-of-a-kind database, unique to Utah that combines many different state and private records related to birth, death and genealogy.
o Very useful for researchers trying to discover genes and hereditary traits.
* CCTS = Center for clinical and translational science(http://www.ncrr.nih.gov/clinical_research_resources/clinical_and_translational_science_awards/)
o A 5-year, $22.5M grant awarded in 2008 from the NIH to accelerate turning basic clinical research into improved patient care.
o Similar grants to 13 other institutions created similar centers around the country.
* HIPAA = Health Insurance Portability and Accountability Act (http://en.wikipedia.org/wiki/HIPPA)
o A set of federal laws that set certain rules related to health information such as patient privacy and confidentiality, information security and transation standards
o This law has impacted many different aspects of health research and clinical processes and procedures. Some were good, some were bad, if you ask 2 people you will get 3 opinions.
* MEDLINE = Medical Literature Analysis and Retrieval System (http://www.ncbi.nlm.nih.gov/pubmed/)
o A massive database (maintained by the NLM) of over 18 million medical articles dating back to the 1950's from about 5,000 journals and publications.
o PubMed is a popular web-based interface for searching the MEDLINE database (also a NLM project)
o Extensively cataloged and indexed to support highly specialized searches, summations and visualizations
+ Uses MeSH (medical subject headings), a large standardized list of article subjects
Approximate top 6 Medical Informatics Programs around the country
* Utah
* Columbia
* Vanderbilt
* Pittsburgh
* Stanford (mostly translational)
* Oregon Health & Science
WORD document of Electronic Health Records (EHR) Terms
www.azdoqit.com/LS4/Glossary%20of%20EHR%20Terms.doc
Week 3 The biomedical world: healthcare systems
Healthcare System
While people often associate the term healthcare with visits to their doctors or time they spent in a hospital, there are numerous aspects to the healthcare system that do not directly involve the care of patients. The people involved in the system include the most obvious participants: patients, physicians, nurses and therapists, as well their supporting staff including nursing assistants, dietitians, pharmacists, social workers, and numerous forms of technologists. Since healthcare is a massive enterprise, there are also administrative personnel such as medical coders, clerks, accountants, computer/information specialists, transcriptionists, quality control specialists, environmental support staff and executives. While not directly involved in patient care, researchers, policy makers, and educators are also part of the healthcare system.
Healthcare is provided to patients in numerous settings including: outpatient clinics, emergency departments, hospitals, skilled nursing facilities ("nursing homes"), and increasingly in the patient's home.
Over the past few decades the quantity and complexity of information used in caring for patients has increased dramatically. With the growth of electronic medical data, a need has arisen for new technologies to manage and organize this data. There are also many emerging opportunities for research to be conducted using this data.
Rising Costs
In the United States, medical costs per capita have been rising exponentially since the 1960's. Compounding this problem is an increase in the number of persons in older demographics, since people tend to consume more health care as they age. In the United States, this increase is due to an increased life expectancy as well as substantial increase in the birth rate after World War II--a demographic often referred to as the "baby-boomers". In this country, healthcare costs have risen more dramatically than in other Western nations. Some of the factors contributing to this are rising costs of pharmaceuticals, increased medical service utilization, and an inefficient payer system.
Despite paying more for their healthcare than in any other nation, Americans often lag behind other industrialized nations on established benchmarks in healthcare. Disparities in the availability of healthcare between the uninsured and insured have led to a general underutilization of services among the uninsured and a likely over utilization of services among the insured. Emergency departments in the US are prohibited from turning people away on the basis of ability to pay. This, in combination with a rising number of uninsured persons has contributed to a huge increase in costs to the system as the uninsured often utilize emergency rooms as a source of primary care. In addition, the uninsured often delay seeking appropriate treatment for relatively minor or chronic illnesses until the problem is more severe and consequently more expensive to treat.
Healthcare Payers
The three major payers in our healthcare system are Medicare, Medicaid, and private insurance companies. Medicare is a government supported healthcare payer for all persons over 65 in the United States. Medicaid provides healthcare for the impoverished and disabled. Private insurance is usually paid for through a combination of individual and employer contributions. Insurance is available on a per person basis for the self-employed, but it is often prohibitively costly.
Health care policy in the United States is heavily influenced by insurance companies, employers, pharmaceutical companies, and many other health care entities (including physician's and nurses groups).
Information Flow in Healthcare Systems
In the past, most medical information was contained in paper charts. Early adopters of computerization in medicine were billing departments and laboratories. These systems were quite limited in their network and data transmission capabilities. Radiology reports and patient tracking systems followed. For each application, hospitals would buy a separate system that best served their purposes each from a different vendor. This was known as the best-of-breed model for software purchasing. The opposing model is where a hospital purchases a one-for-all software package from a single vendor. While this often permits improved connectivity, they are more costly.
Healthcare is undergoing a dramatic change in how services are delivered. In contrast to a patient receiving acute care from a single physician for a defined problem, healthcare has become progressively more focused on maintaining health and treating chronic disease. Today healthcare delivery is more of a team-centered approach where physicians, nurse practitioners, therapists, and social workers all contribute to the development and delivery of the plan of care.
In an ideal healthcare network, hospital information systems would integrate seamlessly with insurance companies and primary care physicians. However, individual physicians in outpatient offices have been slow to adopt integrated information systems. Utah is something of an exception, having two dominant healthcare networks each with an early interest in medical informatics. Integrated information systems are also of interest to government entities and public health researchers. Infectious disease reporting systems and the Utah Genealogy Population Database are currently two recipients of selected types of clinical data.
Another area of active research in medical informatics is the creation of an personal (and portable) electronic medical record. In recent years, commercial companies have taken an interest in this topic with both Microsoft and Google currently working on the creation of online medical record systems. Another concept in development is the creation of implantable id chips and on-person medical records (perhaps on a USB drive?). Many of the barriers to the use of these technologies involve issues of privacy and consumer acceptance. Another barrier is the legacy of hundreds of different systems each with its own standards and protocols.
Food for thought....what are the advantages of paper?
Articles
National Health Information Infrastructure
To provide better health care more efficiently and cost effectively, a study resulted the proposal of a National Health Information Infrastructure. The implementation of this ideal integration and sharing of medical information has been slow. It faces several non-technology related challenges: Privacy, Technology Adoption, and Market Forces:
* Privacy
Concern of safeguarding patient data has led to strict regulating in data usage, maintenance and sharing. Protecting old-faction paper solution of patient records has been relatively easy: lock and key in a secure facility may have proved emotionally sufficient for the public. However when patient data is transmit across the vase public internet, many considered the potential risks outweigh the conveniences.
* Technology Adoption
While mid-size to large-size health care provider had adopted EMR, many smaller primary and outpatient clinics have not. Many of these clinics feel using and maintaining EMR is too difficult or too costly. Their slow adoption creates the bottlenecks in decentralizing patient care from large scale facility towards primary care provider and outpatient clinics.
* Market Forces
To some corporation, sharing patient data may also mean sharing patient profits. They fear they may loose their returning patient. Meanwhile, third party companies, such as Microsoft and Google, are studying ways to provide data management solutions. Eventually, this development may provide the patients more convenient ways to manage their own health and medical records, and to supply the information to the health care provider of their choice when needed.
Additional resources:
* Frontline report on Healthcare Across the World
* Public health reports and statistics for the state of Utah
* US government's webpage for the Medicare system
* The Healthcare and Utilization Project (HCUP) by the Agency for Healthcare Research and Quality
o This links to a series of healthcare databases and software tools for analyzing healthcare statistics.
Kaiser Report on Rising Costs of Healthcare
Week 4 Biomedical data: what is/how it is communicated
Shortliffe Chapter 2
---2.1 What are Medical Data?
Gathering data and interpretation of the data is the central to the health care process. Computers in biomedicine (at the fundamental level) play the role of data collection, storage, and use. This chapter addresses reoccurring set of issues that is pertinent to all aspects of the use of computers in biomedicine, both in the clinical world and in applications related to biology and human genetics.
Data are central to the decision making process. All medical activities involve gathering, analyzing, or using data.
Data
• Provides the bases for categorizing a problem in a patient or population
• Helps the physician decide the flow of further diagnostics
• Provides information that may dictate the most effective treatment
Medical Data cannot be viewed as simply numerical and graphical pieces of information. A variety of subtle types of data may also provide invaluable information to deliver optimal care (for example the patients demeanor during an examination). However, recording these types of data in a meaningful way in a computer of paper record is often difficult.
Datum – is a single observation of a patient (e.g. temperature reading, red blood cell count, blood pressure reading)
Data model – Design of how different types of data are stored and handled by computers. For example blood pressure is viewed as a single datum, but it actually consist of two measurements that are viewed together.
A single datum can generally be viewed with four elements:
1. The patient in question
2. The parameter being observed
3. The value of the parameter in question
4. The time of the observation
It is also important to keep a record of circumstances in which data is taken. These modifiers can provide crucial information for the interpretation of data (Example: blood pressure reading taken after exercise). Uncertainty is also an important component – it is rare that a single observation can be accepted with absolute certainty. The possible response to this is to obtain additional data that confirm or reject the interpretation. There are many reasons why this may not always be feasible (time, cost, delayed treatment). Thus the idea of trade-offs in data collection becomes important in guiding health care decision making.
---2.1.1 What Are the Types of Medical Data?
Types of medical data include:
• Narrative
• Textual data to numerical measurements
• Recorded signals
• Drawings
• Images
Narrative data accounts for a large component. These types of data include the patient’s description of the illness, responses to questions posed by a physician, and social, and family history. This data was traditionally recorded by hand into the medical record, however it more common for these types of data to be dictated and then transcribed via word processor. This allows for more easily integration into the electronic health record.
Narrative data can often include shorthand conventions. This has the benefit of compacting information, however it can lead to confusion for example MI can mean both “mitral insufficiency” and “myocardial infarction.” Many hospitals try to enforce standard for using shorthand.
Numerical data accounts for a large portion of medical data. This often comes in a discrete numerical value, including measurements a laboratory tests to measurements taken during physical examinations. It is important to keep in mind the precision when considering with these types of measurements. For example: is a 1-kg weight fluctuation over a week significant and what external factors can be attributed to the differences (different scales, time of day).
Analog data from continuous measurement pose challenges in how to best store this information in computer systems (Ex. ECG). Visual images are also an important component of the medical record in both sketches by a physician and images, such a radiological.
The storage and subsequent use of the data above is inextricably bound to data recording. The range of data-recording conventions among specialties presents significant challenges to implementing computer-based medical record systems.
---2.1.2 Who Collects the Data?
Physicians are key players in the process of data collection and interpretation. In outpatient and hospital settings, nurses play a central role in making observations and recording data for future reference. Various other health-care workers contribute to the data, including office staff, admission staff, laboratory staff performing tests, pharmacist, and etc. Included in this are the technological devices that generate data.
---2.2 Uses of Medical Data
Medical data are recorded for a variety of purposes. They may be need to support the proper care of the patient from whom they were obtained, but the may contribute to the good of society through the aggregation and analysis of data regarding population of individuals.
---2.2.1 Create the Basis for the Historical Record
Medical records are intended to provide a detailed compilation of information about individual patients:
• What is the patients’ history
• What symptoms has the patient reported? When did they begin, what has seemed to aggravate them, and what has provided relief?
• What physical signs have been noted on examination?
• How have signs and symptoms changed over time?
• What laboratory results have been, or are now available?
• What radiological and other studies have been performed?
• What interventions have been taken?
• What is the reasoning behind those management decisions?
Each new patient compliant and its management can be viewed as a therapeutic experiment, inherently confounded be uncertainty, with the goal of answering three questions:
• What was the nature of the disease or symptom?
• What was the treatment decision?
• What was the outcome of that treatment?
---2.2.2 Support Communication Among Providers
A central function of structured data collection and recording in health care settings is to assist personnel in providing coordinated care to a patient over time. It is less common for a patient to see a single health care provider. The Medical record is important in ensuring quality and continuity of care through adequate recording of details and logic of past conditions and treatment. This is important in health care systems like ours where chronic rather than acute infections increasingly dominate the interactions between patients and their doctors.
---2.2.3 Anticipate Future Health Problems
Medical data are important for screening for risk factors, following a patients’ risk profiles over time, providing a basis for specific patient education or prevention interventions, such as diet, medication, or exercise.
---2.2.4 Record Standard Prevention Measures
The Medical Record serves as a source of data on interventions that have been performed to prevent common or serious disorders. Interventions can include education, immunizations, etc.
---2.2.5 Identify Deviations from Expected Trends
Data often are useful in medicine when viewed as part of a continuum over time. Single data points may have limited use by themselves.
---2.2.6 Provide a Legal Record
The Medical record is a legal document. The medical record is the foundation for determining whether proper care was delivered.
---2.2.7 Support Clinical Research
Medical data used to support clinical research through the aggregation and statistical analysis of observations gathered from populations of patients. Randomized clinical trials (RCT) involve the random assignment of match groups of patients to alternative treatments when there is uncertainty about how best to manage the patient’s problem. Medical knowledge also can be derived from the analysis of large patient data sets even when the patients were not specifically enrolled in an RCT.
---2.3 Weakness of the Traditional Medical Record System
---2.3.1 Pragmatic and Logistical Issues
Data cannot effectively serve health care unless they are recorded and their ability to be used optimally depends of positive responses to the following:
• Can I find the data when needed?
• Can I find the Medical record in which it was recorded?
• Can I find the data within the record?
• Can I find what I need quickly?
• Can I read and interpret the data once I find them?
• Can I update the data reliably with new observations in a form consistent with the requirements for future access by me or other people?
With a paper Medical Record the answer is often no to these questions.
---2.3.2 Redundancy and Inefficiency
Often in a paper medical record information will be recorded redundantly so that summary of important results or information is more easily obtained. Over time this can result in an obese medical record that has the opposite effect making pertinent information harder to obtain.
---2.3.3 Influence on Clinical Research
Clinical research project that require the review of paper charts can be cumbersome, because the researcher must flip through a large amount of information before obtaining the relevant information. Prospective, randomized, double blind studies are considered the best method for determining optimal management of disease.
---2.3.4 The Passive nature of Paper Records
Paper medical records are passive records; they cannot be monitored for their contents unless retrieved manually. This makes them inadequate when medical dogma changes. Conversely an EMR allows constant monitoring in which data can be accessed and then flags and warning can be generated.
---2.4 The Structure of Medical Data
Medicine is remarkable for its failure to develop a standardized vocabulary and nomenclature. Some view this as a problem and others think this is the nature of medicine, and reflects an important distinction between it and the “hard” sciences. This is a major problem when recording data in a computer system, which is most efficiently done when the data are structured and conform to standard syntax.
---2.4.1 Coding Systems
Because of the needs to know about health trends for populations and to recognize epidemics in their early stages, there are various health-reporting requirements for hospitals and practitioners. Another, kind of reporting involves the coding of all discharges diagnosis for hospitalized patients, plus coding of certain procedures that were performed during the hospital stay.
Coding schemes:
• International Classification of Disease (ICD)
• Systemized Nomenclature of Medicine (SNOMED)
• CURRENT Procedural Terminology (CPT)
Health care personnel need standardized terms that can support pooling of data for analysis and can provide criteria for determining charges for individual patients. There is an inherent tension between the need for a coding system that is general enough to cover many different patients and the need for precise and unique terms that accurately apply to a specific patient and do not unduly constrain physicians’ attempts to describe what they observe.
---2.4.2 The Data–to-Knowledge spectrum
Datum – a single observational point that characterizes a relationship
Knowledge – derived through the formal or informal analysis of data
Information – encompasses both organized data and knowledge
Database – a collection of individual observations without any summarizing analysis
Knowledge base – collection of facts, heuristics, and models that can be used for problem solving and analysis of data. Many decision-support systems have been called knowledge-based systems.
---2.5 Strategies of Medical Data Selection and Use
Data collection and interpretation guide the decision making process. This process will differ based of the selectivity that may be applied by one clinician compared to another. These processes of internalization have to be understood to provide better computer-based decision-support tools.
---2.5.1 The Hypothetico-Deductive Approach
The Hypothetico-Deductive Approach is one of sequential, staged data collection, followed by data interpretation and the generation of hypotheses, leading to hypothesis-directed selection of the next most appropriate data to be collected. This process is iterated until one hypothesis reaches a threshold level of certainty. Differential diagnosis comprises the set of possible diagnoses among which the physician must distinguish to determine how best to administer treatment.
---2.5.2 The Relationship Between Data and Hypotheses
Sensitivity – the likelihood that a given datum will be observed in a patient with a given disease or condition
Specificity – an observation is highly specific for a disease if it is generally not seen in patients who do not have a disease
Prevalence – frequency with which the disease occurs in the population of interest
Predictive value (PV) probability that the disease is present based on the results of the test
---2.5.3 Methods for Selecting Questions and Comparing Tests
These issues are covered in chapter 3
---2.6 The Computer and Collection of Medical Data
Data entry has posed a problem for medical-computing since their entry into biomedicine and is probably the number one factor inhibiting the use of computer systems. Several approaches have been used to ease data entry issues. Modern technologies such as wireless networks, handheld devices, and more intuitive operating systems may help to provide the biomedical community with more user-friendly experience.
Kolodner Article------Health Information Technology: Strategic Initiatives, Real Progress
A multifaceted, Balanced Agenda
For more than fifteen years, the adoption and use of electronic health records (EHRs) have been advocated as changes necessary to reduce medical errors and improve the quality and safety of care. Despite efforts of the National Committee on Vital and Health Statistics (NCVHS), health IT progress has been slow To further accelerate the emerging progress, a presidential Executive order issued in April 2004 created the position of national coordinator for health IT. The executive order envisioned better clinical decision-making at the point of care, improved health care quality; reduce health care costs, improved coordination of care and protection of patients’ individually identifiable health information.
“Strategic Framework”-June2004
“Federal Health IT Strategic Plan” –June 2008
Both include broad-spectrum activities necessary to achieve the nationwide, secure, interoperable health IT infrastructure essential for health care transformation.
Recent Activities and Progress
Adoption
The rate of Electronic Health Record (EHR) adoption has been accelerating despite remaining impediments. Adoptions activities include:
1. Establishing and increasing rigorous certification process
2. Develop a valid and reliable survey methodology for measuring EHR adoption
3. Issuing changes to the physician self-referral prohibition and anti-kickback rules
4. Working with malpractice insurance as they developed and implemented plans
5. Launching a Medicare demonstration in twelve communities that provides financial incentives to physicians that use EHRs
6. Establishing in law Medicare payment provision to encourage physicians to use electronic prescribing
Collaborative Governance
Oversight is required at the national, state, and local level to identify health priorities and orders they are addressed. Governance activities have included:
1. Establishing the American Health Information Community (AHIC)
2. Working to establish a public-private entity that can provide shared governance
3. Advancing governance of electronic health information exchange
Privacy and Security
Related governmental activities have included:
1. Working on national level privacy and security policy for the electronic exchange of health information
2. Fostering collaborations within and among more than forty states and territories to review business policies and laws
Interoperability
1. Healthcare Information Technology Standards Panel (HITSP) –harmonizing existing often competing, health IT standards using a transparent consensus based approach.
2. Certification Commission for Healthcare Information Technology (CCHIT) – works these standards into its criteria as it annually increase the requirements for certification of health IT products and services.
The first set of interoperability standards was recognized in January 2008. This includes the Continuity of Care Document (CCD).
The Nationwide Health Information Network (NHIN) will serve as a reliable, secure solution for exchanging electronic health information over the internet. Operation implementation is planned in early 2009.
The Interaction of Technology and Policy
Technologies choices that could influence policy are identified through advocacy groups.
Working Together to Accelerate Progress
Federal advisory bodies such as the NCVHS, will be need for thoughtful investigation and recommendation on heath IT issues. Criticism of the lack of tangible benefits is premature at this point, because the initial standardization only began in 2006.
Interoperability: The Key To The Future Health Care System
Olivier Bodenreider Article: "The United Medical Language System (UMLS): integrating biomedical terminology"
(http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=14681409)
UMLS: What is it? What are the main components of UMLS?
The United Medical Language System is a repository of biomedical vocabularies developed by the US National Library of Medicine. UMLS covers the entire biomedical domain. Major component of the UMLS is the Metathesaurus, a repository of inter-related biomedical concepts. The two other knowledge sources in the UMLS are the Semantic Network, providing high-level categories used to categorize every Metathesaurus concept, and lexical resources including the SPECIALIST lexicon and programs for generating the lexical variants of biomedical terms.
What included in UMLS?
Vocabularies integrated in the UMLS Metathesaurus include the NCBI taxonomy, Gene Ontology, MeSH, OMIM (Online Mendelian Inheritance in Man), the Digital Anatomist Symbolic Knowledge Base, SNOMED (Systematized Nomenclature of Medicine)and others. UMLS concepts are not only inter-related, but may also be linked to external resources such as GenBank.
How UMLS is organized?
In the UMLS, knowledge is organized by concept. Synonymous terms are clustered together to form a concept and concepts are linked to other concepts by means of various types of relationships. Inter-concept relationships are either inherited from the structure of the source vocabularies or generated specifically by the editors of the Metathesaurus. Symbolic relationships can be hierarchical or associative.
What users can do with UMLS?
Users can collect various terms used to name a concept, extract the relations of one concept to other concepts, and obtain a set of concepts for a given category, using the list of concepts that were assigned a given semantic type. UMLS is free (users are required to sign a license agreement).
What tools does UMLS have?
UMLS tools: MetamorphoSys (for customizing the Metathesaurus for customer applications), and program lvg, based on the SPECIALIST lexicon and hand-coded rules (allows users to generate lexical variants).
de Keizer article: "Understanding Terminological Systems I: Terminology and Typology"
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=10786065
Introduction
In this article the term “terminological system” is used as an umbrella term for the notions “classification”, “thesaurus”, “vocabulary”, “nomenclature” and “ontology”. The goal of this article is to provide a referential framework, by which terminological systems can be characterized, understood and compared.
Terminology and Definitions (objects, concepts and designations)
Basic elements used in this article are objects, concepts and designations. "Objects" are "things" ("heart valve") or more abstract things ("pain"). "Concepts" are units of thoughts. Concepts can be described by their characteristics (color (“red”), size (“5 cm2”) and shape (“ellipse”) could describe the concept “ulcer”). Linguistic labels, called terms, are used to designate a concept. Definitions are statements about the meaning of a concept. Codes (letters, numerals or a combination thereof) can be used to designate concepts in a computerized system.
Relationships between Concepts
Concepts can be related to each other by hierarchical [superordinate concept ("heart valve") and a subordinate concept ("mitral valve")]and nonhierarchical relationships. Hierarchical relationships can be generic(“Is_a” relation) or partitive. In the past, generic relationships were called logical relationships in ISO standards. Partitive are relationships in which the superordinate concept denotes an object which represents the whole, and the subordinate concepts refer to its parts (a heart valve is part of the heart).
Definitions of Concepts
The intension of a concept is the set of uniquely describing characteristics including relationships of that concept (set {health problem with anatomical localization “liver”, dysfunction “infection” and etiology “virus”} constitute an intensional definition of the concept “viral hepatitis”). Relationships (and characteristics) can be used to order and define concepts in a system. An intensional definition is a definition based on the intension of the concept. Extensional definition is the set of all specific concepts (species) of a superordinate concept (“Granulocytes” intensionally defined by “Leucocytes with abundant granules in the cytoplasm” or extensionally by the enumeration of “Neutrophil”, “Eosinophil” and “Basophil).
Typology of Terminological Systems
A terminology is a list of terms referring to concepts in a particular domain. A thesaurus is a terminology, in which terms are ordered. When a concept in a terminology or thesaurus is accompanied by a definition, it is called a vocabulary or glossary. A nomenclature is a system of terms composed according to pre-established composition rules or the set of rules itself for composing new complex concepts. Classification is an arrangement of objects or concepts (by the is_member_of relation) based on their essential characteristics into groups of concepts, called classes. A taxonomy is an arrangement of classes according to the "Is_a" relationship from the subordinate class to the superordinate class. A Nosology is a classification of diseases. A terminology, thesaurus, vocabulary, nomenclature or classification is called a coding system when the system uses codes for designating concepts.
Typology of Existing Terminological Systems
Illustrations: types of the terminological systems and the coding schemes used in the terminological systems ICD-9-CM/ICD-10, SNOMED, NHS Clinical terms, UMLS and GALEN.
ICD-9-CM and ICD-10
The International Classification of Diseases (ICD) is maintained by the World Health Organization. The ICD is a classification of generic-related diagnostic terms represented in 17 (ICD-9-CM) or 21 (ICD-10) “chapters”, mainly arranged according to anatomical system or etiology. The codes of the ICD-9-CM and ICD-10 are hierarchical and hence significant.
NHS clinical terms
The NHS clinical terms, formerly called Read Clinical Classification, were originally developed for automated description of clinical and administrative data in general practice, but evolved to a version which facilitates daily care practice in the entire field of health care. It forms a classification of generic-related medical concepts designated by a preferred term and some synonyms (if applicable) which are ordered hierarchically, qualifying it as a thesaurus.
SNOMED
In 1975 the College of American Pathologists published the Systematized Nomenclature of Medicine (SNOMED) to provide terms for a broad range of clinical domains. Its structure consists of eleven modules, also called axes or dimensions (e.g. topography, disease and diagnosis, procedures, etc.) which can be conceived as distinct classifications. Almost all ICD-9-CM terms and codes can be found in the “Disease and Diagnosis” module. By linking terms of the various modules one can compose new complex medical concepts.
UMLS
In 1987 the National Library of Medicine developed the Unified Medical Language System (UMLS). The goal of the UMLS is to facilitate the retrieval and integration of information from multiple machine-readable biomedical information sources such as patient record systems and bibliographic databases. The UMLS consists of four knowledge sources, among which the Metathesaurus and the Semantic network are the most relevant in the context of this paper. As the name indicates, the Metathesaurus is a thesaurus in which concepts are linked to (synonymous and preferred) terms which are alphabetically ordered. The Semantic Network provides information about concepts on a high aggregation level (semantic categories) and their relations. Although in theory semantic relations defined in a semantic network could aid a user to make new composites, in practice the UMLS does not support the composition of new concepts.
GALEN
The goal of the Generalized Architecture for Languages Encyclopedias and Nomenclatures in Medicine (GALEN) project is to formally describe and model the medical domain by which the interchangeability of electronic medical data of different data sources can be supported. The “Terminology Server”, the implementation of GALEN’s goal, integrates three modules: the Concept Module, the Multilingual Module and the Code Conversion module.
Week 5 Biomedical data (metadata/ontologies) & HIS overview
Metadata/Ontologies
Summary of lecture on 9/22/08
* (Most material directly from slides)
* Metadata
o Data about data (Ex: What encoding system used to represent the date?)
o National Library of Medicine resources metadata
o Long definition:
+ “Metadata is all physical data (contained in software and other media) and knowledge (contained in employees and various media) from inside and outside an organization, including information about the physical data, technical and business processes, rules and constraints of the data, and structures of the data used by a corporation.” (Marco, D., Building and Managing the Meta Data Repository: A Full Lifecycle Guide. 2000, New York: JohnWiley & Sons.)
o 3LGM²: Three-layer Graph-based meta model to describe, evaluate and plan health information systems consists of:
+ DOMAIN LAYER (hospital process flow)
+ LOGICAL TOOLS LAYER (HIS information flow)
+ PHYSICAL TOOLS LAYER (HIS information flow)
* Ontology
o Objects in the world with their properties and relations to other objects
o Includes events, processes and states
o Explicit specification of a conceptualization
o Compared to other types of groupings ([Smith, KR-MED 2006],[Chute, JAMIA 2000])
+ Ontology
# Defining types of things and their relations
+ Terminology
# Naming things in a domain
+ Thesaurus
# Organizing things for a given purpose
+ Classification
# Placing things into (arbitrary) classes
+ Knowledge bases
# Asserts knowledge
o Ontology examples:
+ UMLS Semantic Network
# 135 semantic types
# 54 types of relationship
# 6700 semantic relations
+ RxNav
# Browser for RxNorm (NLM repository of standard names for clinical drugs)
o Ontology tools
+ Protege - OWL
* Knowledge Management
o Knowledge
+ Interpreted data, an integration of data, experience and analysis
o Knowledge-based reasoning
+ Experiential-based, data-driven application of knowledge to reason about a clinical situation
o Knowledge management
+ Systems for organizing knowledge; especially to make knowledge computable
+ Ontologies hold this knowledge
+ Ontologies can then be used to compute alerts, info at point of care, etc.
o Reasons for managing biomedical knowledge
+ Large volume of data
+ Volume of data increasing rapidly
+ Allows for quicker delivery of care
+ Allows computation using knowledge
+ Clinicians (especially experts) tend to use knowledge differently, they work backwards from a diagnosis to fit the data
Summary of Bodenreider article
* Introduction
o Discusses three broad applications of biomedical ontologies which constitute the focus of the article
o The ontologies included are:
+ SNOMED CT
+ LOINC (Logical Observation Identifiers, Names, and Codes)
+ FMA (Foundational Model of Anatomy)
+ GO (Gene Ontology)
+ RxNorm
+ NCIT (National Cancer Institute Thesaurus)
+ ICD (International Classification of Diseases)
+ MeSH (Medical Subject Headings)
+ UMLS (Unified Medical Language System)
* 1:Knowledge management
o Annotating data and resources
+ Act as source of vocabulary to annotate documents and data
o Accessing biomedical information
+ Allow us to easily retrieve information, such as through MEDLINE
o Mapping across biomedical ontologies
+ Allow us to identify equivalent terms between ontologies
* 2:Data integration, exchange and semantic interoperability
o Information exchange and semantic interoperability
+ Allow data to be exchanged easily for example between clinical and research settings
o Information and data integration
+ Data can be easily warehoused in a single integrated format
* 3:Decision support and reasoning
o Data selection
+ Groups for studies can easily be determined (Ex: all breast cancer patients) using simple concepts from ontologies
o Data aggregation
+ Identification of traits of certain groups of patients (Ex: long-term cancer survivors) can be easily found
o Decision support
+ Communication to physicians in a standard format is possible, computable knowledge is present, etc.
o Natural Language Processing applications
+ Provide an obvious source of NLP data for things like text mining
o Knowledge discovery
+ Component of data-based biomedical research
Summary of Cimino article
* Introduction
o Talks about increasing importance of biomedical ontologies, the rest of the article will review several currently used ontologies
* Ontology
o Definition
+ "a formal specification of a conceptualization"
* Ontologies and Their Impact
o GALEN (General Architecture for Languages, Encyclopedias and Nomenclatures in Medicine)
o UMLS
o MED (Medical Entities Dictionary)
o SNOMED-CT
o LOINC
o FMA
o GO
o ISO Reference Terminology Model for Nursing Diagnosis
o NDF-RT
o RxNorm
o NCIT
o DOLCE
o Protege (tool)
HIS Overview
Health Information Systems are the major part of the Ideal Healthcare network. It includes many small systems like Admission, Discharge and Transfer summary, Pharmacy, EMR, Billing, Lab, Laboratory Information System, Radiology Information system and many more.
Elements and Components of HIS
The main elements of this system are the clinical , administrative and financial subsystems . To create these subsystems to we need three main essential components
1) hardware : servers; disks, LANs, terminals… (the ITS in University of Utah are involved in providing this component )
2) Software : applications, security, communications… , and more importantly
3) Staff (IT, vendor, clinicians, admin….)
Key functions of the HIS
1. Data acquisition & Presentation,
2. Record Keeping & Access,
3. Communication & Integration of Information,
4. Surveillance and quality measures,
5. Information Storage & Retrieval,
6. Data analysis,
7. Decision support,
8. Education,
9. Research,
10. Quality control / Quality Assurance (QC/QA).
The main problem with the first few functions mentioned involves the security issues associated with handling the data, transferring them across different modules and accessing them. Qualities include both quality of data and the health care system as a whole. Government maintains metrics to evaluate the quality in hospitals at regular intervals which much be met when implementing these systems in hospital environment. The health care system is the broadest kind of information network which allows teaching and training as a part of its development and process.
Options and Factors for Building the HIS
Building these systems involves a lot of discussion and decision about how the goal can be reached. There are many options that are considered when developing/building such systems, which vary from
(1)developing a in-house system from the scratch,
(2)buy a complete Commercial Of The Shelf (COTS) ,
(3)buy ‘best of breed’ modules for the functions needed and put them together and
(4) finally do a mix and match of all the above.
This decision involves a lot of factors namely the functions needed by the hospital, the budget and time available, the size of the hospital (number of beds, number of patients treated everyday, etc) vendor compatibility, decision to follow standards, how much of research involved (which will need a Enterprise Data Warehouse) , etc.
Summary from Group Exercise
To summarize the learning from the group exercise we had in class today it greatly helped to get a good picture of the actual difficulties involved in choosing the right option (in-house, vendor, vendor customized) and the factors affecting decision to build a HIS. The exhaustive list of subsystems demands a very careful understanding of the need of the hospital when building the whole HIS.Not only that building within the given budget and choosing from the long list of subsystems a problem, exercise learning included how much it costs for maintaining such a system in long run which in turn plays a important factor in choosing the type of option. The other considerable problems include, many staff who like what they have now, switching over a real pain, vendor folding their tent at any time, extreme reliability needs.
And finally to mention from observing all the results of the exercise for many subsystems “Vendor” systems will win the deal if our class builds one!!!!
GROUP- ARUP Image:BuildHIS CALC(2).xls
Group Clinicians Image:Wiki BuildHIS CALC.xls
Week 6 HIS case studies
Guest Lecture from Scott Evans with Intermountain Healthcare
1. HELP system is the hospital information system used by Intermountain Healthcare
2. The design of the HELP system was influenced by experience using early computer applications developed in cardiology and intensive care at LDS hospital
1. History --> 35 years old
2. Back when HELP was implemented, almost all hospital systems were administrative or financial. Homer Warner wanted to change that to capture everything that was in the paper chart and put it into an electronic record
3. Warner wanted to be able to do decision support using the record as well as for research.
4. 5 Key features of any good HIS
1. Integrated patient database (all fields need to be coded with a ‘primary time stamp’ on it.
2. There needs to be a knowledge base. This is really just computerized logic or knowledge.
3. Need ability to data- and time-drive the knowledge base
4. Need to be able to use interfaces to buy other ‘best-of-breed’ systems to integrate with what you’ve customized.
5. Everything in the EMR should be stored forever in some retrievable format
3. The EMR is the central hub around which all ancillary systems feed, update, drive and interface to provide clinicians with the ability to better coordinate care.
1. An EMR is great but a longitudinal record nirvana
1. Patient information is now retrievable from anywhere (secured login via VPN too)
2. If a record is longitudinal, you will be able to view any or all of the EMR regardless of the location where care was provided
3. Intermountain has a longitudinal patient record
4. Integrating 3rd party systems with HELP
1. Intermountain purchased the lab and radiology systems that integrate with their HELP system using interfaces
1. Data-driven system within HELP depends on a ‘spooler’, which (post send/receive interfaces) is the electronic waiting room for data, like a print spooler where only one job can be processed at a time, then translated into the correct format and eventually integrated into the medical record for the patient
2. LOINC codes, billing codes, SNOMED, pharmacy data, etc. any kind of data can be passed through this data interface so long as it conforms to an understood and accepted standard
3. Data NEEDS to be CODED!
2. Knowledge bases contain hundreds of thousands of modules of decision support
3. ‘Data Driver’ technology plays critical role in support of EMR
1. An example was given illustrating the role of the data drive process within HELP. See slide on microbiology data going through the data driver ? EMR ?Decision support ? alert file
2. Each coded element (blood for example) does a look-up in the table for coded tables. In the columns associated with blood, then modules for ‘blood’ will run as blood elements come through. The same process holds true for all specimens and their respective elements. There are literally hundreds of thousands of these kinds of modules within HELP.
3. The decision support engine pulls in all pertinent data from EMR depending on the initial logic passed to the decision support engine
4. ‘Time driver’
1. Each HIS that exists has a system clock.
2. Each minute, the time driver looks for MLMs or other clinical programs that may need to be activated. One example is the infectious disease monitor. The alert file is run against the alert file, then against the integrated EMR to find pertinent info, then it prints a report on a scheduled and predetermined basis.
3. It’s like an alarm clock that goes off at a certain time each day, week, etc.
5. Long term storage of database
1. Longitudinal data is a data warehouse ? this is to be kept forever
2. Current database is the CDR
3. Short term archive is not currently used for HELP anymore. However the new ECIS system will have an ODS between the EMR and the data warehouse
4. This type of storage can help shape the future of decision support by discovering trends, processes, etc. that should be captured into logical format
6. Effective decision support
1. Must integrate with daily work process or the clinicians won’t use it!
2. Must be evidenced-based
1. Current
1. Actively used programs are never really complete because they keep changing and they need to be able to adapt so the system can evolve.
2. match local processes of care
3. Easy to use
1. Reliable…if the system is down, it’s worthless
2. Seamless
3. Quick is much better than sexy GUI with bells and whistles.
4. Affordable ? maintenance
5. Reasonable to implement
7. What do we say when people ask what ‘biomedical informaticists’ do?
1. We bridge medicine and IT!
2. What two fields are the most rapidly changing? (IT and medicine… that makes it an difficult but challenging field of work)
8. Infectious Disease Monitor
1. Infectious disease doctors often say “I wish I’d know about this 24 hours earlier”
1. Infections at sterile sites
2. Resistant bacteria strains
3. Notifiable diseases
4. Antibiotic alerts
5. Hospital-acquired infections
2. See slide for the types of systems that are used for the infectious disease monitor
1. Micro, serology, etc…
3. See slide for example of LDSH infectious disease monitor for 25-Aug,1999
1. 3 alerts were kicked off for patient David Jones
4. Over the past 20 years, the ED at LDS hospital has changed thousands of therapeutic decision based on the information contained in the ED infection reports.
9. Comparison of survelillance methods
1. Manually 76% sensitivity and computer was 90% sensitivity with 65% less time required for the computers. The computer sensitivity can improve and must improve and evolve with new logic and enhancements. IHC is now about 98% sensitivity and improving.
10. Computerized surveillance of adverse drug events for hospitalized patients
1. ADE (adverse drug events) reporting program
2. Monitor lab data
1. Data-driven
3. Monitor drug orders
1. Data-driven
4. Monitor drug levels
1. Data-driven
5. Vitals
1. Data-driven
6. Intermountain learned from the Infectious disease control program and adapted lessons learned there to create the ADE program!
7. Using computers to improve and build logic helped discover the magnitude of the ADE problem!
11. Antibiotic Assistant is another tool used at Intermountain. Built on same principles as ADE
1. Patient specific tool
2. CPOE
3. 10 hospitals running this ever day
1. Logic runs back 24 hours, finds a max. Does it again for last 48 hours, then finds max temp for comparison.
4. The antibiotic assistant relies on the knowledge modules and EMR to do real-time prescription analysis for patients as physicians prescribe medications. It makes them aware of any allergies the patient may have along with a knowledge of current meds so as to avoid contraindications.
12. HELP2
1. Web-based in JAVA
2. GUI
1. Can do patient graphs that couldn’t be done in HELP
3. Full of web links to literature
4. EKG data in pdf format is now printable
5. Chest x-rays in picture format available
6. HELP2 has enterprise-wide data available for the entire EMR which is a big improvement over HELP
13. At LDSH
1. Both HELP and HELP2 are used
2. HELP2 was slated to eventually replace HELP. A number of HELP applications and functionalities have been migrated to HELP2 and the new system has some new capabilities and benefits that were not found on the HELP system.
Lecture about UUHSC
* 1. components of UUHSC, UU hospital as the mothership, several large institutes and medical centers, many clinics and community clinics, and a huge telehealth networking system. All through SLC and utah, though not as big as intermountain healthcare system, provide a huge chunk of healthcare in this area.
* 2. What does it take to support such a big system for networking, data flowing? ITS architecture chart is given(http://uuhsc.utah.edu/its/orgchart), provides large operations through departments such as enterprise/networking/data warehouse to keep data center running well, ITS also concerns about education and research.
Citrix as one approach to manage web-based access and applications, The system bases on servers(citrix presentation server and Microsoft windows servers) to support remote access and clinical desktops.
* Wiki’s explanation of (1). Citrix XenApp (formerly Citrix MetaFrame Server and Citrix Presentation Server) is a remote access/application publishing product that allows people to connect to applications available from central servers. One advantage of publishing applications using XenApp is that it lets people connect to these applications remotely, from their homes, airport Internet kiosks, smart phones, and other devices outside of their corporate networks. From an end-user perspective, users can log in to their corporate network from, for example, an airport kiosk, see all of the applications they would see everyday at work, including Outlook email and any internal applications, and access them from the kiosk in a secure environment. To the user, the application would appear as if it was installed and running on their computer (seamless desktop integration), whereas in reality, the application is running on XenApp, usually hosted in their corporate environment.
(2). Thin client: A thin client (sometimes also called a lean or slim client) is a client computer or client software in client-server architecture networks which depends primarily on the central server for processing activities, and mainly focuses on conveying input and output between the user and the remote server. In contrast, a thick or fat client does as much processing as possible and passes only data for communications and storage to the server.
* 3. history of info system development: Lesson from ACIS system—it failed for non-technical reasons, 85% depends on psychological and social reasons, users’ response/human factors have to be taken into account. Now two vendor systems: Cerner, Epic, responsible for inpatient and outpatients settings. Question brought out here: are these two systems integrated together that clinicians can share info across systems?
* 4. Mixed breed products, best-breed modules from a lot of verdors
* 5. Examples from Cerner EMR, different interfaces to show the idea of human factors involeled: need to know how people like to organize data, how does one interface work for people, effective ways to design and evaluate human-computer designation.
* 6. An integrated data model, the middle part provides common services, with this “evidence-based decision support” is possible: example of real-time decision support on the point when a drug like digoxin is ordered, a pop-up window shows notice of “low potassium”, making an evidence-based suggestion. There is a balance between “reminder” and ways to do this, make sure not the dumb way.
* 7. Care transformation project: key is to enhance patient safety. Sometimes it’s hard, like eMAR.
* 8. Care transformation vision: UUHSC’s version of putting hundreds of lines of data into e-Chart.
* 9. Enterprise Data Warehouse: with service layers designed to support interactions with various clinical appplication and knowledge management tools.
Summary of Murff article
Acronyms Defined
* CPRS = Computerized Patient Record System
* CPOE = Computerized Physician Order Entry
* QUIS = Questionnaire for User Interface Satisfaction
Introduction: There are four specific areas in which CPOE shows benefits over traditional paper-based systems for improving patient care. There are:
1. Process improvements
2. Resource utilization
3. Clinical decision support
4. Guideline implementation
Ever though CPOEs can provide the above advantages, in 1998 the percentage of hospitals who required CPOE’s use was still very low. Several concerns can be categorized as follows.
1. CPOE was recommended by national organizations
2. There were a wide choices of available CPOE systems
3. Few studies have evaluated user satisfaction with order entry systems
4. Previous study showed that assessment and incorporation of user feedback concerning info systems is important to ensure proper system utilization
5. There were prior negative experiences with implementation of CPOEs
6. Data showed user satisfaction is an important predictor of s system success
Moreover, Mount Sinai NYU Health Systems train her medicine house staff physicians with two different CPOE systems at two different institutions, this could be ideal setting to study the comparison of the CPOE systems.
1. Hypothesis: different CPOE systems may be not be equally useful, possible reasons were explored.
2. Working hypothesis: with user’s assessments as criteria to evaluate systems’ usability and ID the reasons, comparisons were made between two systems used by the same population of house physician stuff.
Methods adopted in this study include
1. Systems: both systems were mandatorily used and shared similar capabilities, house staff physicians consistently entered the majority of daily orders in both systems. The commercial product is menu-driven with orders entered b mouse, propriety system with a character-based interface, 60-minute training session, CPRS has a graphical user interface with a Microsoft Windows-style event-driven clinical user interface with tab feature, 40-minute training session.
2. Study subjects: anonymous survey by mail, the same population of internal medicine and medicine-pediatrics house staff physicians 144 for commercial product, 132 out of 144 for CPRS
3. Survey administration: QUIS as questionnaire, for commercial product Feb.2000 initial distribution and Mar.3000 reminder, collection ended Apr.2000. For CPRS Apr.2000 initial distribution and reminder one month later, collection ended Jun.2000.
4. Questionnaire of QUIS: a general software-assessment tool, used before for physician satisfaction with electronic medical records. 27-item instrument, 5 categories.
5. Analytic approach: SAS6.12 software for all statistical analysis. Two-tailed unpaired t-tests were used with Bonferroni inequality formula to adjust alpha level and Student-Newman-Keuls test to control multiple comparisons. Comparisons included inexperienced users vs experienced users, light users vs heavy users, recent users vs distant users. ANOVA for post-graduate year level, Spearman correlation coefficient to determine individual questions’ significant relationship with overall satisfaction, univariate and multivariate linear regression to assess demographic characteristics’ influence on overall satisfaction scores.
The results of this study can be generated as follows.
1. Response to survey: 63-64% response rate and 50%-52.5% of gender rate for both systems. Figure 1 Overall level of satisfaction: 0 to 9 range, commercial product had 3.67, CPRS had 7.21. Figure 2 shows for all five categories CPRS had significant higher scores compared to commercial product, especially for “learning” category. Figure 3 shows for every single individual question, CPRS had significant higher scores, the greatest difference was seen for “tasks can be performed in a straight forward manner”.
2. QUIS scores for individual systems: commercial product had only 4 out of 27 questions scored higher than midline value of 4.5, CPRS had all 27 questions higher than midline value.>
3. Correlations of Overall Satisfaction: for commercial system questions associated with “learning the system” are significant, for CPRS questions related to “screen design and layout” are significant. Strong correlation for both systems includes “tasks performed in a straightforward manner”, “remembering names and commands” and “terminology consistency”. One thing important: the question with highest score in commercial product related to “system noise” was not significant. Usage Patterns and Satisfaction: no statistical significant difference between experienced/inexperienced, light/heavy or distant/recent users.
4. Comments Section: commercial product on negative side while CPRS on positive side for patient care. One thing for CPRS, complaint for “system response time”.
Discussion
1. CPRS system is favored compared to commercial system, leading to the conclusion that not all order entry systems are equally usable, caution should be used when making choice.
2. The question “tasks can be performed in a straightforward manner” bears highest absolute difference, together with complaints about “not-intuitive order processing”, showing that the commercial product doesn’t make sense for the location of common orders, some frequently used tests were “buried or nested”. The commercial product also showed improper data sorting routines as “too many irrelevant options were listed on one screen”. On the contrast, CPRS designed eChart with similar line of thinking of a paper chart, which is intuitive and easy for physicians to develop a mental model on expectations/experience.
3. As reported before here it’s also found that user satisfaction correlates best with the ability to perform tasks efficiently, related to (2).
4. One disadvantage of commercial product is that it had a un-common interface.
5. The benefits of CPOEs could be depreciated by bad designations, so continuous collecting/incorporating physician feedback is important.
6. limitations of this study:
1. different amount of rotation times,
2. no assess external factors’ effects,
3. different patient care-mix at the two hospitals,
4. difference on timing of survey administration.
Conclusions: User satisfaction differed significantly between the two order entry systems. Satisfaction was related to user-perceived efficiency in performing necessary tasks. It’s important to ensure physicians’ satisfaction, and also user feedback collection and incorporation in further system setup.
Summary for Saleems’ article:
VHA= Veterans Health Administration
CPRS= Computerized Patient Record System
HCI= human-computer interaction
CRs= clinical reminders
Introduction: The wide implementation of CPRS and its currently being “reengineered” provided a unique opportunity to conduct studies meant to provide prompt feedback to developers to support design changes and enhancements with empirical human factors input. Two design issues focused in the paper are
1. organizing large amount of clinical info in medical record.
2. enhancing usability of CRs. For 1st purpose, “card sorting” a usability approach was used to collect empirical data on how physicians prefer to cognitively organize clinical data when using medical record. For the preliminary phase of 2nd purpose, “simulation study” was conducted to test the design modification’s effect on first-time users’ learnability.
Methods
1. Participants: six resident physicians for “card sorting” as
1. physicians need to access a wider variety of clinical data in CPRS,
2. this 6 had exposure to CPRS but not too much to be biased to either designation. 16 nurses participated in the “simulation study”, they were all experienced with patient check-in/intake, no experience with CPRS.
2. Card sorting: on an empty table, each resident was given a stack of index cards containing data from a comprehensive fictitious patient record and asked to organize the cards into groupings that reflected their preferred organization of the data.
3. Simulation: The current system and the redesigned prototype were both modified in the same fashion, the redesigned system has three modifications. 16 nursing participants were introduced to designs A/B in counter-balanced fashion, time to satisfy a single Pain Screening CR without prior training was recorded(no more than 5min), this time measurement was used as an a priori measure of learnability for first-time user. T-test was used to compare between design A/B.
Results
1. Card Sort: 5/6 grouped data for progress notes, consults, reports and discharge summary, 4 from these 5 created sub-groupings further, the other 1 preferred grouping consults with progress notes. The last one “like to see everything clumped together in a time-based view and then sort from there”.
2. Simulation: time in seconds to satisfy redesigned system was significantly less than time with current system. Discussion
1. Modest changes to user interface in clinical info systems can significantly impact on human performance as illustrated by the “simulation study”, suggesting “tab metaphor” of document types makes sense.
2. Formative assessment of design alternatives with human factors methods such as “card sort” can provide user data to help guide design prior to implementation. “card sort” results in this paper support a customizable view for the “reengineered” CPRS.
3. Pitfalls in this paper: i)Beside ‘card sort” we may need more exhaustive approaches to test on multiple data organization options. ii)nurses’ learning experiment cannot necessarily be extrapolated.
Conclusion: Human factors methods should be routinely used to rapidly collect empirical data to support design decision formatively and throughout the redesign, which can improve user performance and usability as well as reduce cost.
Notes from wiki for a priori:
In statistics, a priori knowledge refers to prior knowledge about a population, rather than that estimated by recent observation. It is common in Bayesian inference to make inferences conditional upon this knowledge, and the integration of a priori knowledge is the central difference between the Bayesian and Frequentist approach to statistics. We need not be 100% certain about something before it can be considered a priori knowledge, but conducting estimation conditional upon assumptions for which there is little evidence should be avoided. A priori knowledge often consists of knowledge of the domain of a parameter (for example, that it is positive) that can be incorporated to improve an estimate. Within this domain the distribution is usually assumed to be uniform in order to take advantage of certain theoretical results (most importantly the central limit theorem).
iii.Data NEEDS to be CODED!
Insert non-formatted text here)
o Additional readings: (your links here...)
o
Week 7 HIS case study and evaluation
October 6, 2008
Introduction to the VA Electronic Medical Records System VistA and its GUI interface CPRS
Presenters and former students of the Department of Biomedical Informatics:
Kevin Meldrum
Curtis Anderson
VA Office of Information Enterprise System Management Office
The search for the Holy Grail…in terms of the Department of Veterans Affairs:
Congress: Stop! Who would dare cross this bridge (over a volcano) must answer me a question, ere the other side he see.
Sir Lancelot (Valiant Developer): Ask me your questions, oh Congress. I am not afraid.
Congress: What... is your quest?
Sir Lancelot (Valiant Developer): To seek the Holy Grail…AND to get physicians to use a computer system to place electronic orders and actually use an electronic health record…AND adopt that system in a relatively short amount of time…WITHOUT being encumbered by excessive bureaucracy…adoption delay…budget cuts…
Congress: You may pass…go on…off you go…
And so forth and so on...
But one must come back across the bridge to get more funding…and is development ever really that easy? And does one always get back across the bridge without being thrown over the edge into the volcano below…? Monty Python humor see: http://www.imdb.com/title/tt0071853/quotes
All of those questions remain to be seen, but one thing that is clear is the Department of Veterans Affairs electronic medical record system, whether one is speaking of the original DHCP system or its renamed infrastructure called VistA and the associated GUI interface called the Computerized Patient Records System (CPRS), has been very successful.
Major distinguishing characteristics of the VA Health Care System?
1) VA is the largest integrated healthcare system in the United States providing care to approximately 5.3 million veterans at 128 medical centers and community based outpatient clinics (over 1400 points of care combined).
2) The VA patient population is characterized by patients that are older, sicker, and poorer. However, these demographics are now rapidly changing with increased numbers of women and younger veterans.
3) Hospital sizes in the VA vary 585,000 Inpatient admissions per year and huge numbers of outpatient visits from 30,000 to 450,000 per year depending on facility.
4) One of the overriding themes of the VA Health Care System then is Data…huge volumes of it. This is largely due to the information system in use and the way in which the VA inputs and stores data.
5) The point of all of this is not about building an IT system, but about results in the form of improved health care quality and health outcomes.
For further reading: VistA Monograph http://www.va.gov/vista_monograph/
Some definitions:
MUMPS(Massachusettes Utility Multi Programming System ) or “M” the underlying programming language of VistA. Some of the strengths of M include its scalability, high productivity, and low hardware requirements. Since it is a hierarchical programming language it is perfect for Medical data. Weaknesses include low transaction reliability, screens that are character based, very few development tools, and poor integrability with other environments. http://en.wikipedia.org/wiki/MUMPS
DHCP(Decentralized Hospital Computer Program) the original no frills, no GUI, character based VA system developed in M as part of a VA “Underground Railway” see: http://www.hardhats.org/history/hardhats.html
VistA(Veterans Integrated Systems Technology Architecture) In 1996 VistA was introduced to describe the automated computer environment that supports daily operations at VA health care facilities replacing DHCP. VistA is an integrated suite of over 100 Health IT Systems used across care settings that includes: Patient Records (CPRS), Order Entry, Ancillary Health, Imaging, Medication Administration, and Data Sharing within and beyond VA facilities.
CPRS(Computerized Patient Records System) the GUI user interface for the VistA system introduced in 1997.
Distinguishing characteristics of the VA IT Infrastructure?
1) In 1982 VA officially recognized and put into place the DHCP system at VA facilities nationwide.
2) Since 2000 the VistA system and CPRS has been in use at nearly all VA facilities.
3) The theme of all VA systems has been local IT control, with local customization and control of CPRS installation at each VA facility.
4) VA IT processing volume is immense. VA uses 128 VistA implementations to provide longitudinal electronic health record services nationwide. The aggregate content of these 128 VistA systems as of March 2007 included 874 Million documents (e.g., Progress Notes, Discharge Summaries, Reports) accumulating at a rate of 640,000 each workday; 1.7 Billion orders (+960,000 each workday); 641 Million images (+913,000 each workday); 1.06 Billion vital sign measurements (+729,000 each workday) and 885 million medication administrations (+620,000 each workday).
5) The VistA Kernel (which includes functions of security, mailman routines, RPC brokering) is central to the system and other database systems and software packages are wrapped around this central core.
6) Problems may arise in the master patient index (MPI) (duplicate SSNs, Vets may have more than one etc, different addresses to reconcile records, multiple birthdays).
7) The MPI allows providers to pull data into one view from facilities where patients previously received health care services.
8) CPRS allows concurrent charting using a package called Text Integration Utilities (TIU) which allows providers to enter free text notes directly into the system during patient care.
9) Standardization already exists within the VA IT infrastructure for example use of HL7 messaging standards and other areas of standardization that have been built for example the VA National Drug File Reference Terminology (VA NDF-RT), or standardization of document note titles and naming conventions.
For Further reading: Steven Brown article on VistA National scale HIS http://www1.va.gov/cprsdemo/docs/VistA_Int_Jrnl_Article.pdf.
VistA/CPRS
VistA/CPRS and its predecessor DHCP, was the fist integrated health record in the country used as a tool for storing data, and delivering patient care and treatment. Also the first electronic medical records system to provide electronic order entry and management, lab results, provider alerts etc. The system also provides a “remote data view” and “VistAWeb” allows users to see health data from other hospitals where the patient has received health services.
VistA Technology Stack (see slide on this in the class presentation)
1) The Backend of VistA is MUMPS (“M”) code. One of the unique things about Mumps is that it has data storage written directly into to it using hierarchical data structures. For example, all variables can have child nodes (called subscripts). Those variables that are prefixed with a “^” (called globals) are run using permanent storage instead of RAM. For a more detailed description of Mumps see the overview found at: http://en.wikipedia.org/wiki/MUMPS
2) VistA Kernel (provides menus, fileman utilities, data dictionaries, scheduling tools etc).
3) VistA core packages (including pharmacy, laboratory, radiology, nursing etc).
4) VistA health systems (including consults, order entry, TIU, imaging etc).
CPRS is highly customizable
1) CPRS has been customized and adapted to meet local and operational needs via local source code modifications.
2) Use of document boilerplates and note templates
3) Ability to implement business rules and order dialogs
4) Customizable hierarchy of parameters
Installation of CPRS Started with one VA site
1) Beginning in 1997 there was gradual adoption of CPRS across VA facilities with the last facility adopting CPRS in 2000.
2) CPRS originally created by people with clinical backgrounds
3) Much of the IT team for CPRS development came from the DBMI at the UofU.
4) National and local leaders were called upon to champion development and facility installations
5) Development teams and users groups were used for iterative development cycles ( LiveMeeting for demonstration)
==Clinical Decision Support in CPRS== (two basic DSS systems in use)
1) Alerting systems (to capture critical lab values for example). This allows real time order checking for drug/drug interactions latest lab results etc.
2) Reminders systems (for example, Influenza vaccines, diabetes management etc).
CPOE in the VA
There is wide discussion and some controversy about CPOE in use at VA Medical Centers and its effects on medication prescribing, medication errors, and adverse drug events.
See a few articles of interest:
1) NY Sun article on CPOE and use of electronic medical records system http://www.nysun.com/opinion/what-the-va-does-right/51962/
2) Article on evaluating clinical decision support systems in the VA http://www.jamia.org/cgi/content/full/15/5/620
3) Special discussion section on CPOE presented in JAMIA. Article http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2215073
4) Article discussing high rates of adverse drug events in a highly computerized hospital environment http://www.informatics-review.com/wiki/index.php/Nebeker_JR%2C_Hoffman_JM%2C_Weir_CR%2C_Bennett_CL%2C_Hurdle_JF._High_rates_of_adverse_drug_events_in_a_highly_computerized_hospital._Arch_Intern_Med._2005_May_23%3B_165%2810%29:_1111-6
Q&A Session:
1) Is CPRS-R (or “second generation CPRS”) Re-engineering Continuing? Originally the direction for next generation CPRS or CPRS-R was to move away from MUMPS toward a Java based system. Congress has since pulled funding for these efforts, but even though it is not formally funded stealth work continues.
2) Is the Electronic Chart as a metaphor still sound? Clinicians can only take steps that are so big…have docs that use CPRS, nurses need to see familiarity with chart on the screen. There is wide demand to go to the next level away from the chart and towards the problem of pulling data sources together and integrated across domain use.
3) How many applications are home grown versus “best of breed”? Most CPRS modules are homegrown. However, with new administrations come new appointees and new priorities. VA is currently in the process of replacing the lab package with Cerner. Most purchased packages tend to be interface systems, lab instrumentation, integration etc.
Demonstration of CPRS
1) The CPRS interface shows all data elements in a medical chart style format. With tabs listing medication orders, laboratory Results, vitals information, or allowing entry and completion of free-text clinical notes.
2) CPRS provides a rich resource of progress notes. Notes documents can be created to include templated sections or other embedded objects like medication orders, laboratory results, vitals information etc gathered from other VistA packages.
3) Each VA site has local committees to set up note templates and formats, and users can build their own custom boilerplate style templates.
4) Each package or module (i.e. laboratory, pharmacy, vitals etc) has a dedicated coordinator. See: http://www1.va.gov/cprsdemo/ for an executable demo of CPRS that you can run yourself.
The future of VistA and CPRS?
VA is moving towards a more service oriented architecture (SOA) delivered via web portals using myHealtheVet see: https://www.myhealth.va.gov/mhv-portal-web/anonymous.portal?_nfpb=true&_nfto=false&_pageLabel=mhvHome. The VA is headed across the proverbial bridge again with recent efforts towards greater standardization, (for example use of SNOMED-CT for many systems), data warehousing, and a more recent initiative towards text-processing and application of natural language processing tools and methods on the huge volume of free-text clinical documents available. In terms of volumes of clinical text, one ongoing study reports that each of the 33,000 surgical patients enrolled had approximately 400 free text documents in text integration utilities (TIU) and a total of 1.5-2.0 megabytes of free-text data in VistA. The infrastructure to support each of these new areas is currently being built.
Additional links, further reading, and papers of interest:
1) Hardhats http://www.hardhats.org/dhcptovista.html description of VA systems development over time and an interesting story.
2) Description of VistA/CPRS http://www.virec.research.va.gov/DataSourcesName/VISTA/VistA.htm
3) VA featured by HHS http://www.hhs.gov/healthit/attachment_2/v.html
4) Computerizing Large Integrated Health Networks By Robert M. Kolodner, Judith V. Douglas http://books.google.com/books?id=zYTTAT7wo-EC&pg=PA137&lpg=PA137&dq=development+of+CPRS+va&source=web&ots=9Qwro2mIjk&sig=c_uyLAAeK9foguo0rBAELFfB9c0&hl=en&sa=X&oi=book_result&resnum=8&ct=result#PPA150,M1
5) VA quality measures see EPRP http://www1.va.gov/vhapublications/ViewPublication.asp?pub_ID=1708 Higher quality diabetes care http://docnews.diabetesjournals.org/cgi/content/full/2/6/4 Fran Weaver article on Vaccinations http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=516244
6) Mary Goldstein articles on ATHENA using the VA EMR see links from: http://healthpolicy.stanford.edu/research/assessment_and_treatment_of_hypertension_evidencebased_automation_athena/
7) Steven Brown NDF-RT Content coverage study http://apelon.com/literature/papers/MEDINFO%202004%20NDFRT%20for%20Mayo%20Drug%20lists%20040204%20Formatted.pdf
A quick review on HIS systems
A hospital information system (HIS) is one computer system which manages the facility’s medical and administrative information allowing clinicians and other health professionals to perform their job tasks.
The HIS focuses on integration of all systems within the hospital, which include:
6. A Clinical Information System (CIS)
7. Financial Information System (FIS)
8. Laboratory Information System (LIS)
9. Nursing Information System (NIS)
10. Pharmacy Information System (PIS)
11. Picture Archiving Communication System (PACS)
12. A Radiology Information System (RIS)
The above systems communicate via DICOM and HL7 information exchange standards.
Definitions
Clinical Information System (CIS) – a system that helps the collection, storage, and manipulation of clinical information. These may be a single specialty, such as laboratory systems or a system wide storage such as the Electronic Medical Record (EMR).
Financial Information System (FIS) – coordinates the fiscal responsibilities of a healthcare facility such as purchasing, payroll, and paying utility bills.
Laboratory Information System (LIS) – manages the laboratory data such as microbiology, hematology, and microbiology.
Nursing Information System (NIS) – an information system using a nursing language such as the North Nursing Intervention Classification (NIC), Nursing Diagnosis Extension and Classification (NDEC)or American Nursing Diagnosis (NANDA) to manage a patients clinical data.
Pharmacy Information System (PIS) – allows the pharmacist to supervise the use of medication in their facility to include activates such as Clinical Screening, Prescription Management, Inventory Management, Patient Drug Profiles, and Report Generation.
Picture Archiving Communication System (PACS) – the system which stores radiologic images.
Radiology Information System (RIS) – the radiology equivalent of the CIS in that it helps in the storage, manipulation and retrieval of radiologic data.
Class Lecture
About HIMSS
The Healthcare Information and Management Systems Society (HIMSS) is the healthcare industry's membership organization exclusively focused on providing global leadership for the optimal use of healthcare information technology (IT) and management systems for the betterment of healthcare. HIMSS frames and leads healthcare public policy and industry practices through its advocacy, educational and professional development initiatives designed to promote information and management systems’ contributions to ensuring quality patient care. (From web site: http://www.himss.org/ASP/aboutHimssHome.asp )
KLAS Quick Facts:
- KLAS is independently owned and operated.
- KLAS' founder is Kent L. Gale.
- The name “KLAS” (pronounced like the word “class”), comes from the first initials of the founder’s names: Kent Gale, Leonard Black, Adam Gale and Scott Holbrook.
- KLAS is headquartered in Orem, Utah, with independent researchers working throughout the country.
- KLAS focuses solely on healthcare technology KLAS is a company that only researches customer satisfaction of HIS vendors. The vendor market includes large and small hospitals, and tertiary centers.
Data collection for evaluations is collected in large enough quantities to provide a reliable data set of a given product. To aid in HIT system acquisition, these surveys provide information on what product works in which situation. For example, the needs of a large hospital HIS system would be different than the small community hospital. Rating the systems is in a color coded format.
Red = bad
Yellow = middle of the road
Green = good
KLAS does not test the actual product in evaluating the systems. Customers polled on the HIS provide the data from either an interview or written survey. The whole corporate philosophy asks ‘What do the people that have to work with these systems think about it’? This company only evaluates vendor systems, none of the Enterprise systems, such as those found at the VA or IMH. Issues found by KLAS include:
• Most systems go in late • Systems around for decades such as Laboratory and Radiology show less satisfaction than new systems. • No vendor does better with inpatient systems than they do for outpatient systems due to the different dynamics of the site.
Computerized Physician Order Entry (CPOE) may develop into a critical link to patient care. The thing is, most hospitals do not use CPOE and those that do have a low use rate by providers. Some reasons the systems have underutilization problems is that it often takes more time to use CPOE than it does to write an order. There is a history with CPOE causing harm to the patient. In the case of a pediatric application, it was due to lack of training of the providers. The consensus is that if the interface and system has a good design it will work well.
User’s Advice for others is an important topic. To get a CPOE system going there has to be an on the floor advocate. The vendor cannot buy their way into the clinical site so they need someone that either mandates the use of the CPOE. The champion has to help sell the system to the staff that does not want to change.
HIS web links
American Medical Informatics Association.
Benchmarking of hospital information systems: Monitoring of discharge letters and scheduling can reveal heterogeneities and time trends. Open Access Research Article. BMC Medical Informatics and Decision Making 2008, 8:15. Accessed at http://www.biomedcentral.com/1472-6947/8/15
EMS Performance Improvement Center- Information on Prehospital Medical Information System
General Electric (GE) HIS site
Healthcare Information and Management Systems Society (HIMSS)
Hospital Information Systems http://www.biohealthmatics.com/technologies/intsys.aspx - Great description of the components of a HIS.
International Medical Informatics Association
KLAS web site
Philips Medical Systems Healthcare Informatics
Siemens Soarian
Week 9 Overview of decision support systems and patient safety
October 20, 2008
Imaging Informatics
o Digital images can be represented and stored in 0’s and 1’s.
o Picture Archiving and Communication Systems (PACS) are computers or networks dedicated to the storage, retrieval, distribution and presentation of images. (Wikipedia)
o Standards include HIS, DICOM, and HL7.
o As technology for imaging informatics advances, image resolution, scanning speed, and image reconstruction time all improve.
Common usage of computers in radiology
o Report generation
o Voice recognition/Transcription
o Electronic signature
o Integration with Hospital Information System (HIS)
o Patient and exam tracking
o World Wide Web and other internet services
o Also in the form of imaging equipments such as CT, MRI, nuclear medicine cameras and digital X-ray systems.
o Teleradiology systems employ computers to acquire, transfer and display images for remote viewing.
CT, MRI and nuclear medicine are the main digital imaging modalities.
o These systems rely on computers to acquire, process, and store image data.
o Information is captured and stored in digital format. When needed, it can be converted to analog format for viewing and printing.
o They require tremendous processing power to generate image data.
o Hardware requirements include large amount of memory, array processors, and powerful graphics processors and libraries.
o Massive amounts of storage are needed for digital images.
Other usage of computers in radiology
o There are a wide variety of systems from many vendors available for patient and exam tracking systems. They are similar or identical to patient tracking in other areas/departments of the hospital.
o Radiology reports could be generated from a typewriter by radiologist, or a dictaphone with transcriptionist processing, or in the most automated setting, a full voice recognition system.
o Patients could use computers to get internet access to the World Wide Web for patient education. They could also use computers to communicate with providers, get results of tests in potentially both text and image formats.
The problems of data overload
o As mentioned before, the resolution of images and number of images are continuously increasing. The types of imaging equipments producing digital images expand from CT and MRI to include ultrasound, digital mammography and digital x-rays. Other sources of digital data arise from electronic communications in radiology between providers as well patient-provider. These all contribute to increase amount of data.
o The massive amounts of data not only create difficulties in data storage, but also in data transfer due to bottlenecks in network speed and capacity.
Imaging Informatics
o It relates to information technology and systems because it affects decision on hardware, software, network, storage technology, and internet.
o It relates to clinical informatics because it is an integral part of medical computing applications and it affects decision making. It also presents unique challenges in data exchange and standardizations.
o It relates to PACS administration in terms of the behavioral aspect, business aspect as well as technical aspect.
o It relates to academics because it is a major theme in research, leadership and education.
Advantages of PACS
o It provides means for easier interpretation and comparison of studies.
o It provides faster and more accurate diagnosis for patients.
o It provides immediate access from any location to comprehensive images and reports.
o It could enhance patient safety through automated process.
o It could facilitate timely remote peer consultation.
o It reduces patient exam delays and report turnaround time, and increases radiology productivity.
o It improves patient/physician satisfaction.
o It makes image enhancement and transmission easier.
o It reduces lost or misplaced films.
o It reduces film and chemical costs, as well as physical storage space requirements.
There are multiple roles of imaging:
o It is used for diagnosis
o It is used for planning and assessment of treatment
o It helps develop guidance of procedures
o It is used in communication of interpretations and findings
o It is used in education and training radiologists/providers
o It is used in research
Radiologic Process
When clinician determines there is a need for radiologic study for a patient, an appropriate procedure would be requested and scheduled. The images acquired from the procedure would be reviewed by a radiologist, who would give the interpretation and create the radiology report. Quality control and workflow monitoring are used for monitoring waiting times, workloads, doses, and complications. Information collected would be used for continuing education and training of radiologists.
Components in Imaging Informatics:
o Image generation: imaging equipment acquires images, digitizing where necessary.
o Image management: this involves storing, transmitting, displaying, retrieving and organizing of images.
o Image manipulation: this includes pre-processing step and post-processing step.
o Image integration: images could be combined with other structured or unstructured data.
Imaging Modalities
o An ideal modality would have high spatial, contrast, and temporal resolution, and at the same time be inexpensive and portable. It should not present risk of morbidity or mortality or ionizing radiation. It should be completely painless and noninvasive. It should provide both anatomic and physiologic information.
o Current modalities include plain film (conventional tomography, fluoroscopy, angiography), cross-sectional imaging (CT, MRI, USG), and nuclear medicine.
o Image parameters: comparison between CR, CT, MRI, USG, and NM.
MRI generates most data per study, and NM generates the least. CR has the highest pixel resolution, and NM has the lowest. CT and MRI both have high contrast resolution, CR is high on spatial resolution, CT and UST are high on temporal resolution. MRI and USG present no radiation. USG is the most portable, while CT is not portable at all. Both MRI and NM provide physiological information. CT and MRI are high on cost, and USG is low on cost.
Requirements for an optimal and functional clinical decision support system
Summary on Kawamoto's paper:(BMJ, doi:10.1136/bmj.38398.500764.8F (published 14 March 2005))
Clinical decision support system (CDSS) definition
Kawamoto and his colleagues defined a clinical decision support system as any electronic or non-electronic system designed to aid directly in clinical decision making, in which characteristics of individual patients are used to generate patient-specific assessments or recommendations that are then presented to clinicians for consideration.
Why the necessity of clinical decision support systems?
Recent research has shown that US adults receive only about half of recommended care, and the US Institute of Medicine has estimated that up to 98 000 US residents die each year as the result of preventable medical errors. To address these deficiencies in care, healthcare organizations are turning to clinical decision support systems. Such systems have been shown to improve prescribing practices, reduce serious medication errors, enhance the delivery of preventive care services, and improve adherence to recommended care standards.
Kawamoto’s study
In their paper, Kawamoto and his colleagues considered 22 potential explanatory features that were identified as being potentially important in the implementation of a clinical decision support system. Such features include general system features, system-clinician interaction features, communication content features, and auxiliary features. Of these 22 features, 15 could be included into their analysis because their presence or absence could be reliably abstracted from most studies under review, whereas the remaining seven could not.
Fifteen features of clinical decision support systems
General system features
1) Integration with charting or order entry system to support workflow integration
2) Use of a computer to generate the decision support
Clinician-system interaction features
3) Automatic provision of decision support as part of clinician workflow
4) No need for additional clinician data entry
5) Request documentation of the reason for not following CDSS recommendations
6) Provision of decision support at time and location of decision making
7) Recommendations executed by noting agreement
Communication content features
8) Provision of a recommendation, not just an assessment
9) Promotion of action rather than inaction
10) Justification of decision support via provision of reasoning
11) Justification of decision support via provision of research evidence
Auxiliary features
12) Local user involvement in development Process
13) Provision of decision support results to patients as well as providers
14) CDSS accompanied by periodic performance feedback
15) CDSS accompanied by conventional education
The seven potential explanatory features of clinical decision support systems
General system features
1)System is fast.
Clinician-system interaction features
2)Saves clinicians time or requires minimal time to use.
3)Clear and intuitive user interface with prominent display of advice.
Communication content features
4)Assessments and recommendations are accurate.
Auxiliary features
5)System developed through iterative refinement process.
6)Alignment of decision support objectives with organisational priorities and with the
beliefs and financial interests of individual clinicians.
7)Active involvement of local opinion leaders .
Main identified types of clinical decision support systems
The commonest types of decision support system were computer based systems that provided patient-specific advice on printed encounter forms or on printouts attached to charts (34%), non-electronic systems that attached patient-specific advice to appropriate charts (26%), and systems that provided decision support within computerised physician order entry systems (16%).
Main features required for an optimal and functional clinical decision support system
Kawamoto's group have identified four features as independent predictors of system effectiveness by a primary meta-regression analysis. The analysis confirmed the critical importance of automatically providing decision support as part of clinician workflow (P < p =" 0.0263)," p =" 0.0187)," p =" 0.0294)." record_id="9728)" system =""> Infobutton can reduce medical errors and cost, and improve healthcare.
What is “Infobutton manager?
“Infobutton manager” is a Independent software component system to make up for the weak points of Infobutton. While many of the information needs arising during typical patient care could be addressed with direct links from the electronic medical record system, infobuttons were not flexible enough to accommodate the variety of information needs that might arise in any given context. Therefore, Prof. Cimino developed Infobutton Manager (IM), that attempts to match the clinician's setting (i.e., the clinician/user's characteristics, the patient's characteristics, the task being performed and the actual patient data available) with the likely information needs and then provides links to resources that can automatically address the need (Trans Am Clin Climatol Assoc. 2007; 118: 273–288)
1) What Infobutton Manager(IM) does:
o To Take “infobutton requests” from clinical information systems
o To provide more-relevant questions that represent the most likely information needs
o To avoid data overwhelming and to minimize cognitive effort
o To avoid navigation workload
=> Advantage of Infobutton Manager: to provide ‘fast and frugal’ decision-making by minimal sufficient knowledge
2) Infobutton Manager deployment
o WebCIS, New York Presbyterian Hospital’s (NYPH) proprietary clinical information system
o New York State Psychiatric System’s PSYCKES system
o Regenstrief Medical Records System at University of Indianapolis
o the NextGen system at Crystal Run Healthcare
Several technical issues have constrained the deployment. For example, integration with the clinical system at New York Presbyterian Hospital required customized programming for each link. A similar experience with MINDscape has been encountered at the University of Washington. The special programming hampered the connections the users' systems to infobuttons
C. Is Infobutton used in a real clinical setting?
"Infobuttons at Intermountain Healthcare, HELP2" (Guilherme Del Fiol et al. AMIA 2006 Symposium Proceedings p.180-184)
o Clinicians at Intermountain have access to a web based EMR called HELP2. HELP2 offers access to a wide variety of data and functions, including laboratory results, clinical notes, problem lists, and medication order entry.
o Infobuttons were first released in HELP2 in September of 2001 in the medication ordering (outpatient), problem list, and laboratory results modules.
o The HELP2 infobuttons use coded clinical data from the Intermountain clinical data repository (CDR) to generate search requests to e-resources. All CDR coded data values can be translated into suitable “free-text” search terms or codes from standard sources, such as ICD-9-CM, LOINC, and the National Drug Codes (NDC). Examples of e-resources currently in use at Intermountain include MDConsult, Clin-eguide Micromedex, UpToDate, and PubMed.
1) Infobutton utilization in IHC (Data from January 1st, 2002 to January 31st, 2006)
Since several limitations were found when using infobutton routines, in 2004, a new software component called E-resources Manager(ERM) was developed to handle all infobutton requests originated from the HELP2 modules. The ERM is composed of four core components: e-resource profiles, e-resource selection, question builder, and query translator.
o Micromedex was the most common resource for medication ordering infobuttons (63% of the sessions), MDConsult the most common for problems (77%), and Clin-eguide the only option for laboratory results. More than half (55%) of the MDConsult sessions were conducted to obtain patient education handouts.
2) Utilization of preference
o The utilization data showed a preference for resources that presented content primarily in summarized format, such as Micromedex, UpToDate, and Clin-eguide. Resources that provide access to the primary literature, such as MDConsult and PubMed, were not as important
D. What are the pros and cons of Infobutton, and evaluation in the articles?
1. Evaluation of Infobuttons(Cimino JJ. AMIA Annu Symp Proc, 2006 , Maviglia SM etal, JAMIA, 2006, 13(1):67-73)
o 74% indicated that infobuttons produced a positive effect on patient care decisions.
o 20% reported positive impact on patient care.
o No reports of negative impact on care.
o Able to meet users’ medication-related information needs in 84%
o Enhanced users’ decisions in 15% of the sessions
2. Pros
o Solution to fulfill clinicians’ information needs at the point-of-care
o Reducing the incidence of serious medical errors due to knowledge gap
o Improving the quality of health care
o Improving patient safety
o Lowering the cost
3. Cons
o High cost of implementation, development, maintenance
o Mapping in the knowledge base are developed and maintained manually.
o Continuous users’ education is recommended.
o Further research is needed to study clinician’s information needs
o Lack of communication standards
Infobutoon Standardization.
Health information systems(HISs) vendors may wish to implement infobuttons in various modules of their applications and such integration is complicated by the lack of a standard application program interface(API), HL7 Clinical Decision Support Technical Committee has been developing a standard for infobutton APIs to support the communication between HISs and IMs, and between IMs and resources.
Arden Syntax Presentation:
Ref. E.H. Shortliffe Biomedical Informatics ch.20 p.705-706, 723-724 / Openclinical
A. What is Arden Syntax
o Standard, formal procedural language that represents medical algorithms in clinical information systems as knowledge modules (Medical Logic Modules, MLMs) (Ref. Openclinical)
o A coding scheme or language that provides a canonical means for writing rules (Medical Logic Modules) that relate specific patient situations to appropriate actions for practitioners to follow. The Arden Syntax standard is maintained by HL7. (Ref. E.H. Shortliffe Biomedical Informatics Ch.20)
1. Background knowledge to help your understanding of Arden Syntax
1) Disadvantages of early Decision Support Systems; not integrated
o Unable to interface with other hospital systems
o Usually require time-consuming manual data entry from the user
o All logic is contained within the code
2) Advantage of integrated Decision Support
o The goal of integrated decision support is error prevention
o Rules are usually simple
o Sent alerts to clinicians warning of potential adverse events
3) The bottleneck of integration
o With the advent of integrated systems, sharing the knowledge was no longer possible
o Difficult to maintain since knowledge is “locked” inside the code
o Not surprisingly, vendors aren’t interested in sharing their code
=> Medical Logic Module (MLM) solves the bottlenecks.
4) What is Medical Logic Module (MLM)?
o A medical logic module (MLM) is a functional unit of a decision support system and an independent unit in a health knowledge base that combines the knowledge required and the definition of the way it should be applied for a single health decision
o MLM is a single chunk of medical reasoning or decision rule, typically encoded using the Arden Syntax.
o MLM has numerous possible functions: Support faster and more accurate medical decision making / Alert clinicians to a potentially dangerous situations / Standardize practice: Quality control and clinical trials
5) The origin of Arden Syntax (MLM and Arden Syntax)
o Reminding of Arden syntax definition again: Arden syntax is standard, formal procedural language that represents medical algorithms in clinical information systems as knowledge modules (Medical Logic Modules, MLMs)
o Example of Medical Logic Module (MLM) written in Arden Syntax; Figure 20.3 in Biomedical Informatics by E.H. Shortliffe
2. History of Arden Syntax development
o Beginning in the 1990s, workers at LDS Hospital, Columbia Presbyterian Medical Center, and elsewhere created
and adopted a standard formalism for encoding decision rules known as the Arden syntax—a programming language that provides a canonical means for writing rules that relate specific patient situations to appropriate actions for practitioners to follow. In the Arden syntax, each decision rule, or HELP sector, is called a medical logic module (MLM).
[What is HELP (Health Evaluation through Logical Processing) system?] Ref. E.H. Shortliffe Biomedical Informatics Ch.20
o HELP system is an integrated hospital information system developed at LDS Hospital in Salt Lake City.
o HELP has the ability to generate alerts when abnormalities in the patient record are noted, and its impact on the development of the field has been immense, with applications and methodologies that span nearly the full range of activities in biomedical informatics.
o HELP adds to a conventional medical-record system a monitoring program and a mechanism for storing decision logic in “HELP sectors” or logic modules. Thus, patient data are available to users who wish to request specific information, and the usual reports and schedules are automatically printed or otherwise communicated by the system.
3. Progress of Arden Syntax
o Arden Syntax for Medical Logic Systems Version 1.0 was adopted by ASTM in 1992
o Version 2.0 was adopted by HL7 and ANSI in August, 1999 (Arden Syntax is now part of HL7, and maintenance of the standard is overseen by the HL7 Arden Syntax Special Interest Group and the Clinical Decision Support Technical Committee.)
o Most recent version is 2.6, approved by HL7 balloting in September 2006
B. What is the basic goal of the Arden Syntax and why is that goal important?
o To develop a system of decision support tools that can be readily shared between institutions
o Designed to be understandable to a person without extensive programming experience
C. Is Arden Syntax used in a real clinical setting?
1. Implementation of Arden Syntax:
o LDS Hospital in Salt Lake City
o Regenstrief Hospital in Ohio
o Linkoping Univeristy in Sweden
o New York Presbyterian Hospital (Developed over 200 MLMs)
2. Vendors who have developed Arden-compliant decision support applications include:
o Eclipsys Corporation
o McKesson Information Solutions
o Siemens Medical Solutions Health Services Corporation
o MICROMEDEX (Medical Logic Modules).
The Arden Syntax makes knowledge portable, but MLMs developed for one environment are not easily embeddable within another. Most commercial applications incorporating MLMs are developed by individual vendors primarily for use within their own environments.
3. Healthcare organizations with Arden-compliant commercial systems:
o Alamance Regional Medical Center, Burlington, NC (Eclipsys)
o Sarasota Memorial Hospital, Sarasota FL (Eclipsys)
o Columbia-Presbyterian Medical Center, New York, NY (developed with IBM)
o JFK Medical Center, Edison, NJ (McKesson)
o Covenant Health, Knoxville, TN (McKesson)
o St. Mary's Hospital, Waterbury, CT (McKesson)
o Mississippi Baptist Health Systems, Jackson, MS (McKesson)
o St. Vincent's Hospital, Birmingham, AL (McKesson)
o St. Mary's Medical Center, Knoxville, TN (McKesson)
o Meridian Health Systems / Jersey Shore Medical Center, Neptune NJ (Siemens)
o Ohio State University, Columbus OH (Siemens)
D. What are the pros and cons of Arden Syntax, and the future of Arden Syntax?
1. Pros (Advantages)
o The target user of the Arden Syntax is the clinician. The Arden Syntax is not a full-feature programming language; for example, it does not include complex structures. MLMs are meant to be written and used by clinicians with little or no programming training.
o Arden provides explicit links to data, triggers events and messages to the target user. It clearly defines the hooks to clinical databases, and defines how an MLM can be called (evoked) from a triggered event.
o The Arden Syntax brings particular support for time functions. Almost all medical knowledge involves the time that something happened. Arden ensures that every data element and every event has a data/time stamp that is clinically significant. Many time functions are provided to help users specify the date and time in MLMs. With any other language, these definitions would be more dependent on the person implementing the MLM; the Arden Syntax allows them to be defined explicitly .
2. Cons (Disadvantages)
o The basic format of Arden Syntax, the MLM, means that it is not the most appropriate format for developing complete electronic guideline applications.
o A problem that occurs with any form of clinical knowledge representation is the need to interact with a clinical database in order to provide alerts and reminders. Database schema, clinical vocabulary and data access methods vary widely so any encoding of clinical knowledge (such as a MLM) must be adapted to the local institution in order to use the local clinical repository. This hinders knowledge sharing. Arden is the only standard for procedurally representating declarative clinical knowledge, so this problem is associated with Arden, but it is not unique to it.
o Another potential limitation of Arden Syntax is that it does not explicitly define notification mechanisms for alerts and reminders. Instead, this is left to local implementation and is, like database queries, contained in curly braces (“{ }”) in a MLM. Explicit notification mechanisms in the Syntax itself may be a part of a future edition.
=> This results in each local implementation having to specify their own interface with their EDW
3. The future of Arden Syntax
o Vocabularies in the Current Arden Syntaxk; Unfortunately, no widespread agreement on standards in this area exists. Nonetheless, candidate vocabularies exist for particular disciplines, such as LOINC for representing concepts concerning laboratory results. Another candidate vocabulary to describe elements such as problem lists, allergies and related items is SNOMED CT
o The continuing issue is to make the Arden syntax compatible with HL7 V3 and the RIM- to make it object oriented.
o Arden Syntax version 3.0 is currently under development by HL7. This version is expected to be based on XML.
E. Summarization (Description) of MLM and Arden Syntax
o The Arden Syntax for Medical Logic Systems encodes medical knowledge in knowledge base form as Medical Logic Modules (MLMs). An MLM is a hybrid between a production rule (i.e. an "if-then" rule) and a procedural formalism. Each MLM is invoked as if it were a single-step "if-then" rule, but then it executes serially as a sequence of instructions, including queries, calculations, logic statements and write statements.
o Arden was developed for embedding MLMs into proprietary clinical information systems. It was designed to support clinical decision making in particular: an individual MLM should contain sufficient logic to make a single medical decision. Sequencing tasks can be modeled by chaining a sequence of MLMs. MLMs have been used to generate clinical alerts and reminders, interpretations, diagnoses, screening for clinical research studies, quality assurance functions, and administrative support.
o With an appropriate computer program (known as an event monitor), MLMs run automatically, generating advice where and when it is needed, e.g. to warn when a patient develops new or worsening kidney failure.
o The initial version of the Arden Syntax was based largely on the encoding scheme for generalized decision support used in the HELP (Health Evaluation through Logical Processing) system for providing alerts and reminders, developed at the LDS hospital in Salt Lake City.
Week 11 Translational informatics and FURTHeR
Lecture Summary
o What is Translational Science?
§ The rate of publications in translational research is growing exponentially. In order to improve the health care system, information needs to move more efficiently through the different facets of the scientific world. Translational science is a growing industry which requires the use of informatics to link three main fields: molecular research, the clinical setting, and the community.
o Misrepresentation of Translational Science
§ The different branches of scientific research are not isolated
§ Individual silos do not exist. Communication among silos is very important
§ The ultimate goal is patient care.
§ Biomedical research at the bench level should also be translated into helping patient care.
o What is Translational Informatics?
§ Basic structure
§ The communication of information is a two way street
§ Bench work <=> Bedside => Community (information is also trafficked back to the bench from the bedside and the community)
§ Bench work <=> Community (Community demographics and studies are essential in research that allows us to focus studies on diseases that may affect a particular population)
§ Role of Informatics
§ Biomedical informatics is focused on moving information efficiently within the health care research system.
§ In the archaic times health care informatics systems included lab data, radiology & path reports, and ADT. The reseach support included Medline, and real-time data acquisition.
§ The current health care informatics systems includes integrated comprehensive EMRs, standards-based technology. In the research front, virtually everything at the bench is automated, the data flow is networked, and there are standard organizations (caBIG).
§ Three key features
§ Information: Need to have a working knowledge of biological pathways, genetic information, and biological knowledge of the clinical problem in hand
§ Computation: Data results are great but analysis is what puts the wheel in motion.
§ Communication: It is important that the different branches of the health care system communicate among each other (Example: FURTHeR). Secondly, scientific knowledge should be accessible to the common man (Example: What GHR does for several genetic disorders).
§ Informatics plays an important role by speeding up the three processes. With the use of information technology we are able to analyze data in minutes and make it available to people around the world.
§ Examples
§ New Work Flows
§ Introduce data repositories based on standard annotations, infrastructures, and services
§ Data pooling and meta-analysis of data
§ Integrated ontologies and knowledge bases
§ Integrating the data that is collected in research studies with clinical data from patients
§ Genomics, Proteomics, Metabolomics...
o How can clinical research be facilitated through translational science?
§ Grid Computing: caBIG
§ Researchers can use EMR data to find potential clinical studies participants
§ Patients can find studies on ClinicalTrials.gov
§ ERICA: Automated IRB System
§ PubMed
§ Google Scholar
Kuhn et al. Summary
The main purpose of the article was to clarify the different challenges in biomedical research and to show how the use of informatics can help move information more efficiently and facilitate the collaboration of bench science, the clinical setting, and the community.
Terms Defined
o Informatics: the main focus of informatics is how information is discovered, created, identified, collected, structured, managed, preserved, accessed, processed, presented, and studied.
o Biomedical informatics: “The scientific field that deals with biomedical information, data, and knowledge - their storage, retrieval, and optimal use for problem solving and decision making."
The Overall Picture: Progress in medicine reflects the collaboration of the two most challenging disciplines that are involved: molecular biology and information technology. Due to the advances in biotechnology and biomedical sciences; a huge amount of data that is available for analysis. The novel findings in molecular biology will have a direct impact on the health care system. Since the completion of the human genome project, a whole new opportunity for biomedical research has been introduced. Now that the human genome has been decoded, researchers are able to find genes that may be prominent in causing various diseases. Similarly the advance in biomedical engineering has resulted in new diagnostic and therapeutic options. Hence the the role of informatics is to create a significant impact in handling and processing huge amounts of information.
o Growth of computing power will allow complex tasks to be solved
o Broadband networks and wireless allows high volume data transfers and sharing of data
o Data volumes will continue to grow; hence, database technology is essential in handling all the data
o Experimentation will become self - organized
o Access to data resources and software tools will become seamless
o Advancement in knowledge management will enable researchers to explore medical knowledge and support knowledge driven discovery
Figure 1: "from single molecules to the entire human population"
o From the perspective of bio sciences, medicine, and public health, informatics plays the role of a critical enabler. For instance, the integration of genetic findings into the clinical environment has an impact on both the clinical research and the clinical practices. Figure 1 portrays the importance of sharing information, and the need for integration and multidisciplinary collaboration. In all these domains, digital data must be stored, processed, and analyzed.
Collaboration and Cooperation
In translational informatics, a multidisciplinary perspective from basic scientific research to clinical medicine and to the community is necessary. Focus in the following areas is important for collaborative research:
o Bioinformatics and systems biology
o Informatics for BME
o Health informatics and eHealth
o Public health informatics
Bioinformatics and Systems Biology
Goal is to understand molecular mechanisms, their genetic framework for diseases and their responsiveness to therapy using advance information technology. With the increase in genomic data, the challenge is to correlate medical phenotype to its genomic and epigenetic counterparts.
Problems
o The transformation of experimental and clinical data into suitable formats
o Software implementation that is acceptable by medical professionals
o Integration of various heterogeneous data sources
o Dealing with large volumes of data sets
Biomedical Engineering
A discipline that advances knowledge in engineering, biology, and medicines, and improves human health through cross-disciplinary activities that integrate the engineering sciences with the biomedical sciences and clinical sciences. In BME, product innovation and development process efficiency are facilitated by informatics and technology.
Problems
o Closer soupling of infomatics and BME is needed
o Emerging challenges such as cognitively adequate real-time visulization, PACS, CBIR
o Handling of streaming data in addition to persistent data
Health informatics and eHealth
eHealth has been defined as the use of information and communication technology (ICT) for health at the local site and froma distance. The driving force behind eHealth is the need for a seamless, high quality, efficient health care system; where, the patient can manage his/her health by improved access to knowledge.
Problems
o Further research on architecture, interoperability, ontologies and standards
o Better data integration and handling of large data sets
o Safe storage and communication of confidential data
o Better workflows
Public Health Informatics
Improving human health is one of the most important goals from a health care perspective. Extensive databases and information platforms combined with powerful technologies will enable the development of new health care tools.
Problems
o Growing demographics
o Integration epidemiology, genetics, clinical medicine, bioinformatics, and informatics
Butte Summary
Translational Bioinformatics
AMIA: The development of storage, analytic, and interpretive methods to optimize the transformation of increasingly voluminous biomedical data into proactive, predictive, preventative, and participatory health.
o Transformation of information from biomedical research into knowledge that can improve human health and disease.
§ First part is to apply scientific research to health care improvements
§ The second part is to take the finding from clinical research and apply it to the health of populations
Eight Reasons why Translational Bioinfomatics is important
o Availability of molecular tools
o Public availability of molecular measurement data
o Culture of sharing molecular data and tools
o Clinicians are expected to interpret bioinformatics methodologies
o Question asking in Translational Bioinformatics
o Calls for translational medicine
o Increasing research funding for Translational Bioinformatics
o Few investigators in Translational Bioinformatics
FURTHeR
What is FURTHeR? Federated Utah Research and Translational Health e-repository. FURTHeR is a large set of componentized data services for the purpose of integrating all health sciences research. It is attempting to provide real time data exchange between large FURTHeR data repository and data service and any number of specific local services (i.e., researchers, patients, health departments, community care providers). FURTHeR is a federated data service, like caBIG (cancer information repository) and the CDC federated service. Everyone is resource constrained, but the FURTHeR projectwill attempt to prove the value of creating and integrating data, with lowest barriers as possible.
What is FURTHeR’s goal:To bring disparate data sources together in order to do research based on this combined and linked sources of information. Data exchange/sharing is the main goal of the FURTHeR domain service. The FURTHeR project includes many services, from query builders for the user, access control services, encryption services and de-identification services, just to name a few.
The FURTHeR project will allow data to be shared among all health related entities, from basic research science to public health institutions and many other related entities.
What will FUTHeR do for you?
1. Enhance your ability to make new discoveries.
2. Change basic research and translate that info clinical practice.
What is the primary source of information?
1. Basic research science
2. Hospitals and clinics
3. Providers and community partners
4. Department of health
What is an example of the FURTHeR concept? A researcher is trying to figure out where to go to get specific information. Currently this researcher must deal with the fact that everyone stores their electronic data in separate containers and in with different structures. For example, birth sex at each institution will store this piece of information in different data formats on different hardware architectures.
How will FURTHeR provide data information to all? FURTHeR will take all data sources, and describe the data of each source. This can be done by creating meta data sources at each site, which describes the structure, relationships, concepts and security of the data that is contained at each of these sites. FURTHeR will have descriptors of data contained at each site so that can queries many sites information. From this understanding about each site’s data, all data will be moved to a single master data repository, known as FURTHeR.
Semantic technologies/ontologies and FURTHeR:The key to the way FURTHeR works is Web 3.0. Web 3.0 uses semantics and anthologies to build user interfaces. Web 3.0 is often referred to as the semantic web. Screens will be built using semantics to define data in a common format known as RDF (an XML formatted “triple store” format of subject, predicate, object.) which can be used describe and relate concepts about each piece of information (similar to object oriented programming concepts). It can take a disease, such as diabetes, and describe it, and relate medications to it, using sematics and ontologies in RDF.
FURTHeR can be used for:
o Initial discovery
o Don’t have to go to each data source, run each query on each source from one location.
o Grant proposal feasibility
Project scoping (do they have enough data to answer my questions).
o And for lower level data analysis
De-identified data (Removed IRB requirements)
Sandbox data
Notes based on Guest Lecturer Scott P. Narus, Biomedical Informatics Dept., University of Utah
Week 12 AMIA, no class
Week 13, part 1: Natural language processing in biomedicine
Problem Set: Mining Clinical and Biomedical Text
Clinical Text
Why Mine Text?
Even though some of the clinical data is properly mapped, using techniques such as controlled vocabulary, large portion of the data still reside in free text, such as discharge notes. These kinds of text contain valuable information such as physicians’ observation, treatment choice, and diagnosis. However, they are also unstructured and high level.
Why is minning the text so difficult?
o The text is unstructured:
Because the text are often compact, concise, and unconventional compared to typical everyday language. The structure is not easily defined.
o The text is also very high level:
They were intended for human and often mainly for documentation purposes.
Biomedical Text
Why mine text?
o Medline’s abstraction maybe mined for research study.
o Full text content maybe mined for knowledge based references.
o Web maybe mined for current trend and professional opinions.
o Email communication may be mined to track epidemic.
Why is mining text in Biomedical Text so difficult?
Like Clinical data, they also contain highly unstructured and high level language only intended for human. Websites, such as blog, and emails contain even lower quality of text.
Answer to mining free text data: Natural Language Process.
To make sense and extract information from a sentence.
Approaches to Natural Language Processing
Statical NLP
o Pro:
§ Adaptable to utilize any machine learning algorithm statistics.
§ Fast. It is often use in search engine
o Cons:
§ The result set often is too high,
§ Accuracy. Contain too many false positive
§ Often workings are difficult to fine tune. The training data set drives the statistic, which drives the behavior of the algorithm.
It is effective in text classification, as a preliminary filter.
Linguistic or Symbolic NLP
o Pro:
§ More accurate then statistic approach.
§ Can extract meaning
§ Open box: Easier to fine tune and correct errors.
o Cons: Speed, often it is slower.
o To develop the algorithm require large number of experts for the targeted text.
It is effective in knowledge extraction.
Statistic NLP:
o Treat text as a bag of words. Each words maybe a potential match as key words.
o Rely on Smart Key words and indexing method to generate a count of a given text.
o Indexing also include statistics of the relationship given a set of words appear within the same document, which is called “n-gram”
However, this method generates too much data to be meaningful for human. There is no meaning attach to the result set, and no interpretation of meaning for each result.
Linguistic / Symbolic NLP:
o Treat text as human would treat them. Parse text into logical units, allowing the algorithm to determine the function and classification of each word.
o Rely on the structural parse tree to give meaning to the words of text:
§ Morphology: Parts of the word, (prefix, suffix)
Words that share similar prefix, suffix and also unusual words such as Kicks, Kick, postnatal, post-surgery
§ Syntax: structural relationships between words (verbs, non, adverbs,
(e.g. I saw a man on the hill with my telescope) Ambiguous
§ Semantic: meaning of the words, phrase and expression (sentence)
(e.g. to be in date for yellow fever)
§ Discourse: The context of the various
§ Pragmatics: The purpose, intent and function of the statement.
§ World Knowledge: Facts and knowledge about the world at large. Common Sense
Current NLP technology can perform relatively effective only to the Semantic level.
Naturally, many solutions are a combination of both approaches.
Application
There are tools that can parse text into Semantic structure and provide meaning to text: MedLee and SemRep
MedLee’s able to parse the following:
“The patient may have a history of MI”
o Problem: my myocardial infarction
o Certainty: moderate
o Status: past history
SemRep
o Symbolic language processing application
o Extract meaning from biomedical literature into semantic prediction.
o Relies on UMLS for knowledge structure.
1. It uses MedLine tiles and other biomedical abstracts as a data set to be used in MedPost, the meaning tagger
2. Also it uses UMLS’s specialized Lexicon to tag each token with lexical tagger.
3. Then Syntactic parser uses the UMLS’s Metathesaurus to parse each meaningful phases creating a MetaMap.
4. SemRep then uses UMLS’s Semantic network MetaMap to make semantic prediction of the text.
The final outcome:
“The leucine-rich repeat kinase 2 mutation is the most common genetic determinant of Parkinson disease…”
“…polymorphisms of PARP-1 contribute to human breast cancer.”
LRRK2 gene PREDISPOSES Parkinson Disease
PARP1 PREDISPOSES Malignant neoplasm of the breast
Some application of SemRep include: automatic summarization, answering clinical questions, literature-based discovery and semi-automated curation.
Week 13, part 2: Nursing Informatics
Lecture Review
What is nursing informatics?
o Nursing informatics (NI) integrates nursing science, computer and information science, and cognitive science to manage, communicate, and expand the data, information, knowledge, and wisdom of nursing practice.--by American Nurses Association
o Discipline-specific area of informatics within healthcare informatics
o Instead of focusing on traditional phenomena of interest to nurses, it focuses on the structure and algorithms of data/ information/ knowledge relevant to nursing practice
o Provides a nursing perspective and illuminates nursing values and beliefs
o According to the American Nurses Association(ANA)(2001)and Staggers and Thompson(2002), NI may be broken down into the following categories: 1) definitions with an information technology focus, 2) conceptually oriented definitions, and 3)definitions that focus on roles.
The evolution of nursing informatics
o 1970s:Starting development of Omaha System; First NANDA meeting
o 1980s:First NI graduate program in Maryland; Graves and Corcoran defined nursing informatics in a seminal paper
o 1990s:First edition of the ANA’s Scope and Standards for Nursing Informatics published; ANCC nursing informatics cerification
Why is nursing informatics so important
o Nurses outnumber any other type of health care provider
o Patients are admitted to tertiary care facilities because they require nursing care, not medical care.
o The nursing workforce possesses unique characteristics that impact informatics solutions.
§ Different levels of entry to practice
§ Diverse educational preparation
§ Different roles in health care settings: Advanced Practice Nurses, Registered Nurses, Licensed Practical Nurses, Certified Nurses Aides, Nursing Assistants and even Family Members
§ Severe and prolonged shortage of Registered Nurses
§ Aging of RN workforce
o Information is a critical component of effective decision-making and high quality nursing practice.
Nurse Decision Making
o Characteristics
§ Consult people, not literature(since literature supporting nurse decision making is unavailable);
§ Intuition and experience highly valued;
§ Immediacy, limited amounts of time for decision making;
§ Complexity of nursing decisions:diverse decision goals and disagreement among decision makers
o Requirement for Computerized Nursing Decision Support
§ A concept-oriented terminology
§ Standardized electronic documentation/ care planning
§ Clinical knowledge models adequate for decision support
§ Identification of appropriate semi-structured decisions
§ Ease of use
§ Integration into workflow
§ Mobile/ point of care
Nursing Terminology
o Structured Terminology is Important:Evidence based nursing practice; Decision Support; Research; Interoperability; Productivity and efficiency
o Possible Terminologies for Nursing Diagnose/Problem
§ SNOMED CT
§ Clinical terminology comprised of codes, terms and relationships
§ Concept-based: Each code represents a single meaning and can have multiple descriptions (terms)
§ Vertical structure:Parent child relationships
§ Horizontal relationships:Relationships between concepts
§ North American Nursing Diagnosis Association: NANDA-I Nursing Diagnosis Development
§ Develop, refine and promote terminology that accurately reflects nurses' clinical judgments
§ Six Axes: 1)Diagnostic Concept 2)Acuity 3)Unit of Care 4)Developmental Stage 5) Potentiality 6) Descriptor
§ Twelve Domains: Health Perception/Health Management; Nutrition...
§ Multi-axial approach
§ International Classification of Nursing Practice (ICNP)
§ A unified nursing language system
§ Compositional terminology facilitating cross-mapping of local terms and existing terminologies
§ Clinical Care Classification (CCC)
§ Standardized, coded nursing terminology that identifies the discrete elements of nursing practice.
§ Consists of two interrelated terminologies - the CCC of Nursing Diagnoses and Outcomes and the CCC of Nursing Interventions and Actions
§ Omaha System
§ Perioperative Nursing Data Set (PNDS)
o Possible Terminologies for Nursing Intervention/Order
§ SNOMED CT
§ Nursing Interventsion Classification (NIC)
§ NIC is a standardized language for treatments that nurses perform.
§ The Classification includes the interventions that nurses do on behalf of patients.
§ An intervention is defined as "any treatment, based upon clinical judgment and knowledge, that a nurse performs to enhance patient/client outcomes."
§ It was developed at the University of Iowa
§ Clinical Care Classification (CCC)
§ Omaha System
§ Perioperative Nursing Data Set (PNDS)
§ Logical Observations Identifiers Names and Codes (LOINC)
§ A six axis model:
Example:
3140-1 BODY SURF AREA PT ^PATIENT QN DERIVED
o Possible Terminologies for Nursing Outcomes
§ SNOMED CT
§ Nursing Outcomes Classification (NOC)
§ Clinical Care Classification (CCC)
§ Omaha System
§ Perioperative Nursing Data Set (PNDS)
o Comparisons of Nursing Standardized Terminologies
Nursing Minimum Data Set Systems
o Identify essential, common, and core data elements to be collected for all patients/ clients receiving care
o Based on U.S. Hospital Discharge Data Set
o U.S. NMDS Data Elements
§ Nursing Care Elements(nursing diagnosis, nursing intervention, nursing outcome, and intensity of nursing care)
§ Patient or client demographic elements( personal identification, date of birth, sex, race and ethnicity, residence)
§ Service Elements(unique facility or service agency number, unique health record number or patient or client, unique number of principle registered nurse provider, episode admission or encounter date, discharge or termination date, disposition of patient or client, expected payer for most of this bill)
o Other MDS
§ NMMDS (nursing management minimum data set)
§ iNMDS: provides a framework for collecting information to describe and examine nursing practice,nursing resources and selected healthcare problems.
Summary for Ozbolt on a brief history of nursing informatics
o Started in the second half of the 20th century,
o 1950's
§ Nurse researcher were asked by IBM to provide consultation about possible uses of computers in health care.
o 1960's
§ the American Nurses’ Association appointed a committee to identify priorities for investigation: a focus on nurses’ use of information in communicating and decision-making was included
o 1970's
§ First reports of “computer applications in nursing” began to appear
§ Nursing care planning systems: to relieve the burden of documentation and to improve the quality and completeness of the plan.
§ Technicon Medical Information System(first comprehensive hospital information system) at El Camino Hospital integrated nursing care planning, documentation, and feedback
§ Federal agencies supporting projects to advance nursing informatics:
§ National Center for Health Services Research (NCHSR):healthcare information systems including nursing care planning and documentation.
§ Department of Defense: Tri-Service Medical Information System (TRIMIS)
§ Veterans Health Administration:Start to create its own clinical medical record system
§ Division of Nursing (DN): a series grants to Omaha System
§ Nurses involved in interdisciplinary efforts to develop and implement health care information systems: contribution a nursing perspective to the first CPOE, etc.
§ Nursing Knowledge representation and decision support:University of Michigan developed prototypes to formulate nursing diagnoses from assessment data
§ No definitions of nursing diagnosis in the literature
§ North American Nursing Diagnosis Association held its first meeting: identifying an initial set of 37 nursing diagnoses.
o 1980's
§ Introduction of the personal computer in 1980 facilitated small-scale prototyping of nursing informatics applications.
§ Introduced informatics courses in schools of nursing
§ A working conference convened by Werley defined 16 data elements to be collected on all patients: four uniquely nursing elements were nursing diagnosis, nursing intervention, nursing outcome, and intensity of nursing care.
§ Nursing informatics as a distinct specialty in nursing: American Nurses Association and the National League for Nursing respectively established a Council and a Forum on Computer Applications in Nursing.
§ First graduate education program in nursing informatics opened at University of Mariland; Nursing informatics was defined as a scientific discipline
§ Judy Ozbolt convened an expert panel to identify priorities for research in the field: establishment of data standards was fundamental to unleashing the potential of nursing informatics to improve practice.
o 1990's
§ Technological advances: Internet, Web-based applications, Laptops, PDA, MEDLINE
§ An invitational working conference: Next Generation Nursing Information Systems: Essential Characteristics for Professional Practice, described attributes that would support nursing care delivery and documentation, quality improvement,and nursing research.
§ Nursing Informatics Working Group (NI-WG) of AMIA was formed
§ American Nurses Association published the first versions of the Scope of Nursing Informatics Practice and the Standards of Nursing Informatics Practice.
§ Additional graduate programs were established at the University of Utah,the University of Colorado, Duke University, and etc. Nancy Staggers launched a research program on nurse-computer interaction.
§ Standards of Nursing Language and Terminology:
§ Nursing Interventions Lexicon and Taxonomy
§ Home Health Care Classification (HHCC,was renamed Clinical Care Classification System.)
§ International Classification of Nursing Practice (ICNP).
§ Integrate nursing concepts into SNOMED.
§ An invitational working conference convened by Judy Ozbolt agreed to develop concept-oriented reference terminology models for nursing.
o 2000's
o Emerging Developments
o Summary
§ the lack of standards for language and data limited the usefulness of early applications in Nursing Informatics.
§ How to achieve computability and semantic interoperability--Collaboration across disciplines and national boundaries
§ The necessity to realize the potential of nurses to transform and improve health care and outcomes through informatics
Summary for Bakken et al on the future of nursing informatics
Anursing informatics research agenda for 2008–18: Contextual influences and key components
o 3 aspects of context that influence Nursing Informatics research:
§ 1)genomic health care: incorporation and use of genomic data into nursing research
§ 2)shifting research paradigms: translation research
§ 3) social (Web 2.0) technologies
o Key components of a nursing informatics research agenda for the next decade (2008 –18)
§ User Information Needs: data tsunami
§ Acquisition, Representation, and Storage of Data, Information, and Knowledge: acquisition, representation, and storage of genomic and environmental data in a manner that supports visualization and analysis in conjunction with patient and nursing data is needed
§ Informatics Support for Nursing and Healthcare Practice:the re-engineering of nursing practice within the context of interdisciplinary care teams
§ Informatics Support for Patients/Consumers and Families: develope and apply informatics strategies enabled by Web 2.0 and future technologies to empower patients and their caregivers for collaborative knowledge development
§ Informatics Support for Practice-based Knowledge Generation: support knowledge generation from practice.
§ Design and Evaluation Methodologies: a broader conceptualization of evaluation across stages of the system development life cycle and the need to match evaluation method to stage of development.
o Conclusion: nursing informatics agenda for 2008–18 must:
§ expand users of interest to include interdisciplinary researchers;
§ build upon the knowledge gained in nursing concept representation to address genomic and environmental data;
§ guide the reengineering of nursing practice;
§ harness new technologies to empower patients and their caregivers for collaborative knowledge development;
§ develop userconfigurable software approaches that support complex data visualization, analysis, and predictive modeling;
§ facilitate the development of middle-range nursing informatics theories;
§ encourage innovative evaluation methodologies that attend to human-computer interface factors and organizational context.
Additional Resources
o Interview with Carol Romano
o American Nursing Informatics Association
o Some Nursing Informatics Defintions
o National Institute of Nursing Research
o SNOMED Clinical Terms User Guide
o Nursing Management Minimum Data Set (NMMDS)
o The Nursing Minimum Data Set: abstraction tool for standardized, comparable, essential data.
Week 14, part 1: Informatics and Cognitive Science
Cognitive Structure • Patterns of association comprise a “mental representation” and are constantly changing
Cognitive Function • Thinking and action consists of spreading “activation” across associative networks • Some linkages, such as habits, are so strong that activation across associative structure occurs instantly • Awareness is related to the “slowness” of the activation
Associative Processing • Activation of one part of aschema results in a sense of expectation for the other parts-attention is directed • For example, if one expects an apple to be red, it is likely to be seen and remembered that way • Takes intense cognitive effort (yoga, meditation) to avoid schematic processing
Human Error and Cognition• Slips: behavior is captured by the wrong cognitive structure (likely to occur for a well learned behavior) so that one drives home when wanting to stop at the store • Lapses: goal is dropped in process (forgetting what you are doing mid-task) • Mistakes: wrong goal • Most human errors are related to slips and lapses
Theoretical Implications Attentional resources are limited • Bank of neurons related to awareness is small • Interruptions interfere with closure of a task
Dual processes operate continuously (Associative processing vs. symbolic processing)• Associative processing: gradual accretion of knowledge through progressive associations, fast/effortless thinking, slow change, no awareness required, resistant to impact of cognitive load • Symbolic memory processing: fast increase in knowledge, slow/effortful knowledge, awareness required, fast change, highly sensitive to cognitive load • Implications: human preference to minimize cognitive load, incorporate adaptive strategies to think less • Motivation: increased cognitive load (work, distractions) associated with greater work by associative system (less thinking, greater pattern recognition processing); increased motivation for accuracy associated with greater work by symbolic system
Effects of Effortful Processing• Example: Two groups of MDs; one group given “simple” scenario that included all necessary information; other group had to search for required information; results differed significantly • Hypothesis that greater time searching for information led to a more well thought out decision, generation of additional hypotheses
Information Overload• Mismatch between available attention and demands in the context • Associated with disorientation, inability to determine relevance, distraction, high effort, lack situational awareness, inability to think • Results due to inattention to work processes in the design/implementation process • For example, deviations in work-flow are perceived as interruptions and changes in information timing increases cognitive effort
Information Processing is Goal Based Thinking is Goal Directed • Perception: do not see items outside goals • Attention: focus • Recall: memory organized around goals
Motivation and Emotion • Anxiety narrows perceptual field • Interest focuses attention • Theory states that we forget failures, overestimate successes, desire to maintain control • Therefore, consider designing web interfaces to draw interest
Knowledge Management Goals • Minimize cognitive load: write lists as reminders, delegate, eliminate task • To be accurate: find knowledge sources, slow down to think, identify redundant/confirming sources • To be fast: short-cuts, minimize steps
Problem of Work-Arounds • Change work processes, use adaptive strategies to avoid information overload
Levels of Expert Control• Skill-based: sensory/motor cues activate attention and behavior (for example, stopping at a stop sign), requires few cognitive resources, assessed by observation • Rule-based: memorized rule, required some cognitive effort, assessed by interviews, surveys, observation • Knowledge-based: different plans considered, simulated outcomes compared to goals, highly effortful thinking, assessed by interviews, decision-analysis
Human-Computer Interaction (HCI) Concepts Human Factors • Applied computer science and psychology • Definition: scientific study of the interaction between people, machines, and their work environments • Examples: color coding classes of drugs, interruptions in medication safety
Ergonomics• Focuses on issues of human safety, convenience, and function in relationship to machines, tools, and equipment people use to perform jobs • Example: monitor location in workstation
HCI • Definition: study of how people design, implement, and use interactive computer systems, and how these systems affect individuals, organizations, and society • Staggers HCI Framework and HCI goals (effectiveness, efficiency, safety) detailed in article summary (below)
Usability• Definition: extent to which a product can be used by specific users in a specific context to achieve specific goals with effectiveness, efficiency, and satisfaction
• Although the concepts of “usability,” “user satisfaction,” and “usefulness” are different, these terminologies have often been used interchangeably.In the discussion of system acceptability, Jakob Nielsen describes the concept of usefulness, which includes both usability and utility. Usability of a system is associated with how well users can use that functionality. On the other hand, utility of a system is associated with whether the functionality of the system can do what is needed. Nielsen describes user satisfaction as an attribute of usability. When assessing outcomes of CIS implementation, however, this conceptualization may need modification.
Usability Problems • Complaints: Cerner difficult to use • De-installation of clinical systems: Sharp Hospital (lack fit with workflow, separate financial/clinical systems, different user interfaces)
DoD Usability Problems• Failed first user acceptance test, designed for primary care but deployed to all providers so does not work for sub-specialties, lacked lab interface interoperability, decreased physician productivity, averaged slow response time between modules
Axioms of Usability• Early and central focus on users • Iterative design of applications, empirical measurement (effectiveness, efficiency, satisfaction) • Web design example: improved interface by adding graphics, easy visual representation of potential problems, differences in icons/color to provide immediate cues to potential problems
Resources for Interested Clinicians • Siemens, IDX clinical systems, Microsoft, Bell Labs, Intel, BMI 6820 Human Systems Interaction Course
Article Summary: "Human-Computer Interaction in Health Care Organizations"
Purpose of HCI and Usability
1. To promote acceptance of systems by creating better software and developing improved applications to support specific work
1. The Problem of Unusable Systems:
1. System usability is a major determinant of acceptance or rejection of a health care information system
2. Example 1: Pilot study found that installation of a integrated clinical workstation was unsuccessful due to the following:
1. user customization was not sufficiently robust;
2. system integration was insufficient; and 3) poor presentation of data
3. Example 2: The US Department of Defense Composite Health Care System II failed user acceptability testing for the following reasons:
1. web-based application was too slow to support clinical processes; and
2. the notes function was considered cumbersome by clinicians. Also, clinical users were not included in the design and development of the product (Goedert, 2000)
3. Usability issues can result in productivity declines, treatment delays, user frustration, underutilization of applications, errors, need for additional personnel, and additional funding to address redesign issues
2. Potential Benefits of Incorporating Human-Computer Interaction Concepts into Health Care Information Systems
1. Potential benefits include the following: 1) patient safety, due in part to correct data entry, display, and interpretation for sound decision making; 2) reduction in time to complete tasks, implement training, and conduct software rewrites
2. Significant cost savings from implementing usability techniques are reported
3. Mayo Clinic usability laboratory, shared Medical Systems, and Cerner are among the few usability initiatives in health care
3. Definition of Terms
1. Human Factors: scientific study of the interaction between people, machines and their work environments (Beard & Peterson, 1989)
2. Human-Computer Interaction (HCI): study of how people design, implement, and use interactive computer systems and how these systems affect individuals, organizations, and society (Myers, Hollan, & Cruz, 1996)
3. Ergonomics: focuses on the design and implementation of equipment, tools, and machines in relation to human safety, comfort, and convenience (Loegendoen & Costa, 1994)
4. Usability: addresses specific issues of human performance during computer interactions within a particular context (Rubin, 1994). The field incorporates topics such as the following: ease of using an application, ease of learning, ease of remembering interaction methods, user satisfaction with system use, efficiency of use, error-free/error-forgiving interactions, and seamless fit with workflow. Usability is aided by designing systems that require a minimum of learning and are easy to remember
4. Goals of Human-Computer Interaction and Usability Concepts
1. HCI goals: promote acceptance and use of systems by creating better interactive systems and better software, developing new kinds of applications to support work, and promoting job optimization with the use of information systems
2. HCI goals may be expressed by overall effectiveness, efficiency, and satisfaction of users’ interactions with information systems (Fig. 15-1)
3. Effectiveness relates to completeness of needed functions
4. Application safety is critical in life-critical systems (for example, ICU)
5. Efficiency relates to the expenditure of resources, which include time, energy, error rates of users, and cost to the organization (support personnel, application redesigns)
6. HCI focus:
1. interactions of the user and computer; or
2. work activities of group of users (computer-supported cooperative work)
5. Axioms of Usability
i. An early and central focus on users in the design and development of systems
ii. An iterative design of applications
iii. Empirical usability measures or observations of users and information systems
The process includes the following:
. Users evaluate early prototypes to assess effectiveness and efficiency
a. Usability problems are sent to designers for correction
b. Users re-evaluate corrected prototypes
c. Usability test includes structured observations of users and computers to define the problem areas for redesign and system impact on costs and productivity (both user and organization)
d. Off the shelf systems: require customization to local users, tasks, environment
A Framework for HCI in Health Care Contexts The Staggers Health HCI framework provides a model with the following elements: patients, providers, computers, context, tasks, information, interactions, and a development trajectory over time (Fig. 15-2)
Patient, Provider, and Computer Behaviors
. Push technology: selected data are automatically delivered to user’s computer
i. Pull technology: user requests information by performing a search
ii. User interface: allows humans to computers to cooperatively perform tasks, usually via a computer screen
Characteristics of Patients, Providers, and Computers
. Humans: attributes such as gender, educational level, computer experience
i. Computers: attributes related to specific hardware and software
The Task Information Exchange Process
Each specific task is associated with a goal for task completion (exploratory or detailed plan)
Health Context
1. Context can include the following: setting (physical or virtual), environment (e.g., corporate structure), individual or corporate cultures, social norms, and physical features (e.g., lighting or noise)
1. Informatics Development Trajectory
1. Humans and computers progress along a development trajectory
2. Early interactions: less information on screen may be preferred to users
3. Later interactions: greater screen density preferred over clicking multiple screens
2. Performing Usability Assessments
1. Usability assessment: includes structured usability testing, as well as other techniques, such task analysis to determine functional requirements
2. Type of assessment depends on when the assessment is targeted in the system’s life-cycle
3. Five Types of Usability Assessments
1. Usability assessment to determine users’ needs and requirements
1. Completed at the beginning of system’s life cycle: survey users’ characteristics, task activities, and interactions among users and tasks in specific environments
2. Exploratory Test
1. Conducted after determination of preliminary designs
2. Major goal is to assess the effectiveness of emerging design concepts
3. Assessment Test
1. Conducted early or midway into application development. Evaluates lower level operations (efficiency) and task presentation
4. Validation Test
1. Performed late in the development cycle
2. Assesses how well all of the modules in an application work as integrated whole
3. Includes benchmark standards
5. Comparison Test
1. Conducted at any point in a system’s life cycle
2. Determines which application is easier to learn, effective, etc
4. Usability Indicators (Table 15-1)
5. Sample Methods in Usability Assessments
1. Task analysis: use of systematic methods to determine what users are required to do with systems by accounting for behavioral actions between users and computers
2. Informatics specialists: 1) record user actions in flowcharts and task descriptions; and 2) assess critical tasks to assist in system selection and purchase
3. Think aloud technique: users talk about what they are doing as they interact with an application
4. Cognitive walk through: detailed review of real/proposed actions to complete a task in a system
5. Heuristic evaluations: assessments of a product according to usability guidelines, such as Web design principles (Box 15-3)
6. Usability questionnaires, such as the Questionnaire for User Interaction Satisfaction (QUIS) (Normal et al. 1998), Purdue Usability Testing Questionnaire (Lin, Choong, & Salvendy, 1997), and Software Usability Measurement Inventory (Kirakowski & Corbett, 1993)
7. Ethnographic Techniques: methodology in which a researcher describes the subject’s point of view with a focus on the experience in social settings
8. Contextual Inquiry: methodology that involves observing representative users in actual work settings
6. Steps in Conducting Usability Assessments
1. Step 1: Define the purpose
2. Step 2: Assess constraints (time, user expertise, software availability)
3. Step 3: Use the HCI framework to refine each component Sample questions: Which users? What about the tasks are of interest?
4. Step 4: Emphasize some components
5. Step 5: Match methods to the purpose, constraints, and framework assessment results
Additional readings:
Human Factors and Ergonomics Society
HCI Literature
Top 10 Web Design Errors
Week 14, part 2: Informatics in Clinical Education
INFORMATICS AND MEDICAL STUDENTS: In order to improve the quality of care provided by the future physicians the medical students of today need to learn the use of computer based tools and Information management in the delivery of healthcare. Which will train the clinicians on major focus areas like: EMR and their use, Evidence Based Medicine, Coding, Decision support systems and the use of technology in there are of expertise.
* INFORMATICS AND TRAINING OF HEALTHCARE PROFESSIONAL: It has been an area of major concern since early eighties by various leaders in Informatics at that time, US organizations, Association of American Medical College and International Medical Informatics association. This led to the initiative of adding informatics to the medical school curriculum. At present 86% of medical schools include Medical Informatics as a required course. (AAMC, 2008 LCME Part II Annual Medical School Questionnaire)
The U’SOM is revising its curriculum and plans to incorporate medical informatics in the ‘Clinical Medicine’ component. Also to enhance the overall use of technology by using computer based exercises and simulation. Thus the BMI department was invited to suggest additional informatics content for incorporation into the new curriculum.
* OBJECTIVE:
In her lecture, Dr. Beaudoin presented the approach used by the Department to determine what additional informatics knowledge and skills should be included in the new curriculum. The goal of the BMI Department is to develop a plan to integrate informatics competencies into the medical school curriculum.
* CURRICULUM:
CURRENT CURRICULUM BY MS YEAR:
MS I: Orientation to online resources/pub med, Introduction to biostatistics, clinical epidemiology, medical literature analysis
MS II: Review of evidence-based resources, Focus on EBM skills
MS III: Focus on case-based clinical problem solving and literature review (Ob/Gyn, Peds, Family Medicine)
MS IV: Biomedical Informatics elective
PROPOSED CURRICULUM IN PHASES:
PHASE I: Introduction to informatics tools, including orientation to radiology output, EMR architecture,search strategies, EBM skills, etc. Also included is a list of quantitative methods related to population health, biostatistics, epidemiology and research design, Mainly Medical Sciences, little focus on Medical arts and clinical Medicine
PHASE II: Mainly Medical sciences: Molecules, cells and cancer, Host and defense, Life cycle, Metabolism and Reproduction, Circulation and Respiration, Brain and Behavior, Skin bones and Joints, , and a little focus Medical Arts and clinical Medicine(Family practice and Ambulatory Clerkship)
PHASE III: Mainly Clinical Medicine (clinical clerkships) with a little focus on Medical Sciences and Medical Arts
PHASE IV: Equal focus on Clinical medicine, Medical science and Medical arts
* INPUTS TO THE PROCESS FROM THE CURRENT TO PROPOSED CURRICULUM:
- Faculty Focus Groups: Considering the future of medical practice
- Faculty Approach Nation Wide: at Florida State University, University of Colorado, College of Medicine, Phoenix.Integration of medical informatics throughout the process
- AAMC Recommendation: Medical school Objective project (MSOP), identified five roles for physicians: Lifelong learner, clinician, - Educator/Communicator, researcher, Manager
- Student Input: Previous surveys of students and residents on perceived competencies on computer knowledge and skill and Informatics knowledge and skills. In order to gather perspectives about the importance of acquiring informatics competencies in terms of their future careers. A voluntary web based survey was taken by the U medical students in all the four years.
* APPROACH BY DEPARTMENT OF BMI IN THE DEVELOPMENT OF CURRICULUM:
Map content onto curriculum- Implement Curriculum- Assess student progress-Create /revise objectives
* PHASE I- MEETING I
TEAMS INVOLVED: DBMI, College of Pharmacy, Eccles Library, SOM (2 MS)
AGENDA:
* Reviewed MSOP learning objectives
* Life-long learner, clinician, educator/communicator
* Revised objectives, suggested new domains
* Developed specific recommendations:
- Informatics concepts should be separated from the technology
- Informatics learning objectives related to Continuous Professional Development should be included in the curriculum, and other domains considered such as Consumer Health, Genomics and Personalized Medicine and Public Health,among others
- An introduction to informatics ‘basics’ should be provided to medical students
- Applied informatics should be integrated into the new curriculum
- Orientation to/instruction in the use of the electronic medical record systems used by the University Hospital,the VA Hospital, IMC and other clinical rotation sites should be provided prior to the start of medical student rotations
- Emphasis should be placed on a team-based approach to clinical care
- The following implementation strategy to incorporate informatics content should be considered:
- Phase 1: Introduction to basics, Phase 11 and Phase 111: Integration, Phase 1V: Informatics elective and/or BMI Certificateprogram
- Solicited input from BMI faculty/others
- Shared recommendations with CTC
* PHASE I- MEETING II
TEAMS INVOLVED: DBMI, College of Pharmacy, Eccles Library
AGENDA
- Reviewed MSOP learning objectives
Life-long learner, Researcher, Clinician, Educator, Manager and User of Technology…look up the slides for details on specific recommendations
- Revised objectives, suggested new role...Physician as ‘User of Technology'
- Solicited input from BMI faculty/others
- Finalized list of 25 objectives...To be shared with Office of Curriculum and Medical Education
* PHASE II: Implementation
* EDUCATIONAL INITIATIVES:
- AMIA informatics specialty certification
- Development of core content for physicians
- Future adaptation for non-clinicians
- AMIA “10 x 10 Program”
- AMIA partnerships
- AHIMA: Define core competencies for those working with EHRs
- AAHC: Develop multidisciplinary informatics curriculum for clinical students
- TIGER Initiative: Enable nurses to engage in digital era of health care
- CDC: public health informatics training
* PERSONAL EXPERIENCE AS A MEDICAL STUDENT:
Based on my experience I think it is an excellent idea to expose the medical students to the area of clinical informatics, may be a few courses in MS I year on the basics of Informatics, clinical systems, EMR, CPOE, CPRSS etc . But due to the overwhelming amount of information that the medical students have to study, I think it should be "optional" to take extra courses, electives, rotations or even a certificate for the students who are interested in this area. I have learned at AMIA talking to various Physician Informatics that there have been talks between the AMIA academic forum and American Board of Medical Specialties to make “Clinical Informatics” a medical specialty so the medical students/physicians(like me) who are interested in this area have a lot of opportunities to come…..
* Personal experience from a clinical instructor in Sonography and Radiography.
I agree with Neelam that medical professionals, physicians, nurses or allied health, need to know the basics of biomedical informatics. The big problem is that there is not enough time to cover all the basics much less something new. The only way to include more is to increase the time people are in school but how can we do that? Right now allied health education is maxed out on credits as per our accredited guidelines. As we saw in Nursing education, there is a hierarchy in medical education within the professions. Perhaps these advanced topics, such as biomedical theory, would have to be part of a BS or higher degree.
* Additional links:
AAMC:
AMIA acdemic forum
AAMC, LCME medical school questionaire
Jerant AF, Lloyd AJ. Fam. Med. 2000 Apr;32(4):267-72:
AHIMA
AAHC
The Technology Informatics Guiding Education Reform(TIGER)Initiative. The TIGER Initiative emerged from a national gathering of leaders from nursing administration, practice, education, informatics, technology, and government, as well as other key stakeholders, who realized that nursing must transform itself as a profession to realize the benefits that electronic patient records can provide ( HIMSS Nursing Task Force 2007; Sensmeier 2007; TIGER 2007 )
*
Week 15 Special topic presentations
Institutional Review Board (IRB)
What is the IRB?
* Defined by the Research Act of 1974
* Mandated in the US by 45 CFR (Code of Federal Regulations) Part 46
* Assures, both in advance and by periodic review, that appropriate steps are taken to protect the rights and welfare of human research subjects
Human subjects institutional review boards(IRBs) were created as a direct result of ethical concerns about the preservation of autonomy, beneficence,nonmaleficence, and justice pertaining to research in human subjects.
1. Safeguards are provided to protect the rights and welfare of study subjects.
2. Risks to the study subjects do not outweigh benefits or knowledge gained through the study.
3. Selection of study subjects is fair and equitable.
4. Each study subject gives informed consent prior to participation, according to FDA requirements.
5. Adequate evaluation of the data ensures the ongoing safety of study subjects.
6. Provisions are made to protect the privacy of study subjects, and the confidentiality of data collected.
http://www.copernicusgroup.com/tcg/whatis.php
Instances of Medical Abuse
* Physical abuse:Tuskegee Syphilis Study, Nazi human experimentation
* Psychological abuse:Stanford prison experiment, Milgram experiment
* Psychological and physical abuse: MK Ultra Project
Belmont ReportBased on National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research (1974-1978), the Department of Health, Education and Welfare (HEW) and was a result of teh Tuskegee Study. “Ethical Principles and Guidelines for the Protection of Human Subjects of Research” published in 1978 and called the Belmont Report because the first draft of the report began in the Belmont Conference Center. Created the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research. Lists ethical principles for human subjects in research. (1) respect for persons: protecting the autonomy of study subjects and treating them with courtesy and respect and allowing for informed consent; (2) beneficence: maximizing benefits for the research project while minimizing risks to the research subjects; and (3) justice: ensuring reasonable, non-exploitative, and well-considered procedures are administered fairly (the fair distribution of costs and benefits.)
How this applies to an IRB • Reference for research involving human subjects.
• Ensures ethical research
Objectives
* federal mandates to protect subjects through the detailed process.
* While this process is basically the same for every board, it is customized for each institution.
o Institutional Review Board (IRB) for U of U found at http://www.bmi.utah.edu/?pageId=2312
+ The required documents for the U of U IRB. http://www.research.utah.edu/irb/submissions/pdf/Required_Documents_for_BM-Jul08.pdf
+ An U of U IRB is required for a study involving any of the following institutions:University of Utah Main Campus, University of Utah Health Sciences Center, Associated Regional & University Pathologists (ARUP),Eccles Institute of Human Genetics, Huntsman Cancer Institute,Moran Eye Center, All center and clinics managed by University Health Care, Primary Children's Medical Center, VA Salt Lake City Medical Center, Shriners Hospital for Children, Salt Lake City
* There are separate forms for Intermountain Health System and U of U.Intermountain Healthcare IRB site http://intermountainhealthcare.org/xp/public/research/irb/
IRB Process
* Submitted by the Principle Investigator (PI)
* The fees
* The complete IRB submission contains:
1. an institution-specific application for research protocols.
2. a protocol narrative includes the purpose, hypotheses, procedures, risks, benefits, options, financial interest, authorization under HIPAA, etc.
3. consent forms regarding the research complexity and the needs of the population being studied.
Health Insurance Portability and Accountability Act (HIPAA)
On February 16, 2006, HHS (Department of Health and Human Services) issued the Final Rule regarding HIPAA enforcement. It became effective on March 16, 2006. The Enforcement Rule sets civil money penalties for violating HIPAA rules and establishes procedures for investigations and hearings for HIPAA violations. Title I regulates the availability and breadth of group health plans and certain individual health insurance policies, such as Medicare and Medicaid. It amended the Employee Retirement Income Security Act, the Public Health Service Act, and the Internal Revenue Code protecting health insurance coverage for workers and their families when they change or lose their jobs. Title II of HIPAA, known as the Administrative Simplification (AS) provisions, requires the establishment of national standards for electronic health care transactions and national identifiers for providers, health insurance plans, and employers as well as the security and privacy of health data.
The Privacy Rule took effect on April 14, 2003. It establishes regulations for the use and disclosure of Protected Health Information (PHI). PHI is any information about health status, provision of health care, or payment for health care that can be linked to an individual. It requires covered entities to notify individuals of uses of their PHI.
The Security Rule complements the Privacy Rule. While the Privacy Rule pertains to all Protected Health Information (PHI) including paper and electronic, the Security Rule deals specifically with Electronic Protected Health Information (EPHI). It lays out three types of security safeguards required for compliance: administrative, physical, and technical. For each of these types, the Rule identifies various security standards, and for each standard, it names both required and addressable implementation specifications. Required specifications must be adopted and administered as dictated by the Rule.
Portability and Accountability
* Portability: it protects individuals from losing their health insurance when leaving and/or changing jobs by providing insurance continuity
* Accountability: it increases the federal government's authority over fraud and abuse in the health care arena
Consent - Clinical Informed consent forms for research studies now are required to include extensive detail on how the participant's protected health information will be kept private, making the already complex legalistic section on privacy documents even less user-friendly for patients who are asked to read and sign them.
* Section 164.58
o Only covered entity can access information
o Consent for others required
* Researcher cannot contact subject
* Subject can revoke consent at any time
HIPAA / IRB Waiver of Authorization requirement Granted by IRB
* Little or no risk to subject
* De-identified data
* Use of a limited data set
* Group size too large to obtain consent
* Inability to conduct research
Consent vs authorization
* Consent – oral or written agreement to participate in a clinical trial
* Authorization – permission for release of information.
Why is HIPAA important?
* HIPAA provides increased protection for patient privacy as patients themselves need to authorize the release of their health information.
* With the development of the Electronic Health Record (EHR), as we moved patient information to the electronic medium, developed integrated systems across the continuum of care, and released and redisclosed information to many people and agencies which needed access to the information, standardized federal legislation became an imperative.
* HIPAA was designed to guarantee that information, transferred from one facility to the next, would be protected.
* In an electronic environment, protecting privacy has become extremely difficult, and patients are becoming increasingly concerned about the loss of privacy and the inability to control the dissemination of the information about them.
Impact of HIPAA at work
* the cost of Health Care:. In the short term, rather than reducing costs, there have been increased administrative costs and complexities for clinical practice.
* the complexities educating of Employees and Patients: . HIPAA obligates the employers to educate the staff, and it obligates all of us to inform patients about privacy.
* implications for homeland Security and Disaster Planning: Health care providers will be required to balance the need to treat patients safely with the need of the government to gain access to information in order to protect citizens.
* issues related to a unique Patient Identifier:Development of this unique identifier could be one of the most contentious and controversial aspects of HIPAA.
* the compilation of personal health records: American Health Information Management Association (AHIMA)website MyPHR Personal Health Record ( www.myphr.com ) gives patients and providers information about HIPAA, confidentiality, patients' rights with regards to their health information, and important links for additional information to support the electronic health records.
* the impact on research initiatives: Based on HIPAA, patient authorization is required for research, unless a waiver has been granted by an IRB or Privacy Board and the standards require de-identification of data for research purposes.
Implementation at the U of U setting
* ERICA (Electronic Research Integrity & Compliance Administration):https://erica.research.utah.edu/erica/Rooms/DisplayPages/LayoutInitial?Container=com.webridge.entity.Entity%5BOID%5B5FD2DA60262617429607E459C0E09D92%5D%5D
Pros of HIPAA
* increased patient privacy and protections against fraud and abuse
* increased the ability to share patient information across the continuum of health care
* The standardization and security of health information is essential for homeland security and access to personal information will be necessary in crises, such as epidemics, terrorist attacks, or natural disasters.
* the assignment of a unique patient identifier(one of the goals of HIPAA) will facilitate link patient information across a continuum of care, which can reduce medical errors.
Cons of HIPAA
* increased the cost and complexity of health care in the short term
* increased patients' concerns about privacy
* continued employee's frustration about the implementation of the HIPAA rules and regulations
* HIPAA is not the panacea for solving all the problems related to privacy
New terms generated by HIPAA
* Covered entities (CE): include health care providers, health plans, and health care clearinghouses that are compelled to protect individually identifiable health information for oral, paper, and electronic communication.
* Protected health information (PHI): is individually identifiable health information relating to an individual's past, present, or future physical or mental health condition, provision of health care, or payment for the provision of health care. It also includes names, telephone numbers, addresses, medical record numbers, and Social Security numbers (Office of Civil Rights, 2005).
* Designated record sets: include information such as medical and billing records. However, peer review documents, appointment and surgery schedules, or employer records would not be considered as part of a designated record set.
Additional reading
* US Department of Health & Human Services http://www.hhs.gov/ocr/hipaa/
* http://privacy.med.miami.edu/glossary/xd_protected_health_info.htm
* Reference: Neamatullah, I., Douglass, M., Lehman, L., et al. Automated de-identification of free-text medical records. BMC Medical Informatics and Decision Making. 2008, 8:32. This article is available from: http://biomedcentral.com/i472-1472-6947/8/32
* NIH web site http://privacyruleandresearch.nih.gov/pr_08.asp
* Health Insurance Portability and Accountability Act Privacy Rule Causes Ongoing Concerns among Clinicians and Researchers by Jennifer Fisher Wilson located at web site http://www.annals.org/content/vol145/issue4/
* For more information on Sec 164.506 visit http://frwebgate.access.gpo.gov/cgi-bin/get-cfr.cgi?TITLE=45&PART=164&SECTION=506&TYPE=TEXT or http://privacy.med.miami.edu/glossary/xd_consent.htm
* UU consent forms web page http://www.research.utah.edu/irb/forms/hipaa/index.html
* NIH site http://privacyruleandresearch.nih.gov/clin_research.asp
* Institutional Review Boards and the HIPAA Privacy Rule http://privacyruleandresearch.nih.gov/irbandprivacyrule.asp
* Office of NIH History http://history.nih.gov/01docs/historical/2020b.htm
* The President’s Council on Bioethics http://www.bioethics.gov/reports/past_commissions/index.html
Challenges of De-identifying Free-text Clinical Notes
What is De-identification?
* 1996 Congress passes Health Insurance Portability and Accountability Act (HIPAA), Calls for standards to protect health information (PHI).
* “Statistical proof” – PHI cannot be linked to an individual, “Records are de-identified when the risk is very small that the information can be used alone or in combination with other reasonably available information to re-identify the individual”. Statistical proof is fairly vague and there is the question of how this would be applied to free-text clinical note documents and other types of electronic documents.
* “Safe Harbor” – requires complete removal (scrubbing) of 18 types of PHI to be considered “de-identified”
HIPAA “Safe Harbor” Identifiers(*not applicable to text reports)
* Names
* geographic subdivisions smaller than a state (street address, city, county, precinct)
* For dates directly related to the individual, all elements of dates, except year. (date of birth, admission date, discharge date, date of death)
* All ages over 89 or dates indicating such an age
* Telephone number
* Fax numbers
* Email address
* Social Security Number
* Medical Record Number
* Health Plan Number
* Account numbers
* Certificate or license numbers
* Vehicle identification/serial numbers including license plate numbers
* Device identification/serial numbers
* Universal Resource Locaators (URL’s)
* Internet Protocol addresses (IP’s)
* Biometric identifiers*
* Full face photographs and comparable images*
Why is de-identification necessary?
* Increasing amount of research utilizes clinical documents
o De-identified documents could be used to refine research questions/feasibility, develop pilot studies, leverage available EMR data
o Research guided by ethical principles of beneficence (and nonmaleficence)
Methods to remove PHI from clinical documents
* Manual – involves human review
* Automated – involves machines and NLP
How are these methods evaluated?
* Evaluated by standard statistical measures of accuracy. Recall is more important for the specific use case of de-identification.
* Precision and Recall are two widely used measures for evaluating the quality of results in domains such as Information Retrieval and statistical classification. In a statistical classification task, the Precision for a class is the number of true positives (i.e. the number of items correctly labeled as belonging to the class) divided by the total number of elements labeled as belonging to the class (i.e. the sum of true positives and false positives, which are items incorrectly labeled as belonging to the class). Recall in this context is defined as the number of true positives divided by the total number of elements that actually belong to the class (i.e. the sum of true positives and false negatives, which are items which were not labeled as belonging to that class but should have been).
Image:Testoutcome.JPG
·
o Recall (sensitivity): TP / (TP + FN)
o Precision (PPV): P = TP / (TP + FP)
o F-measure: Fb = (1+ b2 ) ´ P ´ R / (b2 ´ P + R)
Measurement of performance
* Failure to remove any one of the HIPAA “safe harbor” PHI data elements (undermarking) - %FN (1-Sensitivity)
* Removing too much information than is required (overmarking) - %FP (1-Specificity)
* How well the method does at removing only relevant information - measured by PPV (precision)
A reliable reference standard is the cornerstone of system evaluation
Must develop reference standards to train and test NLP systems, $65/hr trained nurse abstractor manual review and annotation, Domain expertise (physicians, nurses, nurse practitioners, pharmacists, physician assistants) does the annotation task depends on the use case and annotation goals. Annotation tool(s)
* Many tools available to build reference standards: GATE, Knowtator, iAnnotate, Callisto, oXygen, XMLSpy , etc.
* Knowtator
o Open source
o has specific features useful for this type of work : Protégé plugin, easily create annotation schemas, import/export data as XML, calculate Inter-annotator agreement (IAA)
Manual vs Automated
* Manual De-identification
o Time cost for manual review and removal of PHI is high
o Manual review/removal requires efforts of more than one reviewer to improve recall (sensitivity).
o Problem for manual review is FN cases.
o Manual methods using simple search and replace functions do not perform as well as human review
o Published performance manual methods(Douglas et al. and others): Still have 2% or more FN for one use case; With 3 people reviewing documents for PHI = better Recall, but, even with 3 trained human reviewers still have 1-2% FN!
* Automated NLP Methods
Replaces PHI with specific tags using offsets and proxies. Originally developed for pathology reports, expanded to clinical documents . Uses heuristics, rule sets, lexical and syntactic context. Applies dictionaries of geographic locations, hospital names, popular names found in the U.S. Census. Uses the UMLS Metathesaurus® to preserve medical words or phrases
* Commercial Software:
o De-IData http://www.de-idata.com
+ Developed at University of Pittsburgh Medical Center
+ In use by University of Pittsburgh Medical Center, Vanderbilt University
o DeIdentify http://www.deidentify.com
+ Leads to Google Health instead
+ Did Google buy this?
o MedLEE (Friedman, Columbia University)
+ not designed to identify PHI
+ dates, names, gender, age can be extracted from the output
* Developed using research funding:
o Scrub (Sweeney, MIT) http://privacy.cs.cmu.edu/people/sweeney/scrub.html
o MEDTAG (Ruch et al., Switzerland)
o HMS Scrubber (Beckwith et al., Harvard)
o MeDS (Friedlin et al., Regenstrief)
o Stat De-ID (Uzuner et al., University of Albany)
o deid (Neamatullah et al., MIT) http://www.physionet.org/physiotools/deid
Published performance automated NLP methods: The best of these still 1% FN!
Three specific examples of Automated NLP Methods
1. Medical De-identification System (MeDS)( Developed by Jeff Friedlin (currently at Reigenstrief) and Clem MacDonald while Jeff was an NLM Fellow)
* Input: raw free-text clinical document
* Output: same (but processed) free-text clinical document with PHI removed
o Removes 18 HIPAA-specified identifiers
o Removes provider information such as names, office phone and fax numbers, office addresses, and health care institution names
2. Medical Language Extraction and Encoding System(MedLEE): General purpose NLP system, not originally designed for the task of de-identification
* Input: free-text clinical notes
* Output: XML output of medical concepts assigned to semantic categories and concept modifiers: structured data coded using UMLS
o Potential for PHI removal?
o Could be used to enhance other de-identification systems
3. i2b2 Challenge: Integrating informatics for biology and the bedside
* i2b2 de-identification challenge http://www.i2b2.org/de-id/demo.php (MIT, SUNY)
o NLP challenge to remove PHI from discharge summaries
o 889 records “de-identified” and “re-identified” with realistic surrogates
o Records enriched with ambiguities of PHI with non-PHI
o Data enriched with out-of-vocabulary PHI surrogates (i.e., made up names)
o 669 records for training and 220 for testing
* 7 teams submitted 16 systems in total
* Common methods used: BIO model, POS tagging, N-grams, Orthography, Templates / regular expressions, Machine learning (CRF, SVM, rules, boosted c4.5)
* Overall, statistical learning systems with regular expression template features performed best
Conclusion
* Must have reference standard for evaluation: Reference standards are use case specific
* Manual approaches high cost, impractical for large scale application
* Automated approaches may not be generalizable , must be trained and tested for a specific use case
* De-ID software had the highest published sensitivity (recall) (99.3%) and specificity (94.9%).
o Software cost De-ID for one license $50,000
* Question: How good is good enough?
Further reading
* Neamatullah I. Automated De-Identification of Free-Text Medical Records. MIT Dept of EECS, MEng thesis, 2006.
* Douglass M. Computer-Assisted De-identification of Free-text Nursing Notes. MIT Dept of EECS, MEng thesis, 2005.
* Levine JM. De-identification of ICU patient records. MIT Dept of EECS, MEng thesis, 2003.
* Meystre S, Savova G, Kipper-Schuler K, Hurdle JF. Extracting Information from Textual Documents in the Electronic Health Record: A Review of Recent Research IMIA Yearbook of Medical Informatics. 2008:138-54.
* i2b2 (Informatics for Intergating Biology and the Bedside) website. [cited 11/30/2008]; Available from: https://www.i2b2.org/.
Semantic Webs,Semantic Networks, and A Use Case(Cleveland Clinic's application)
What is the Semantic Web?
It is an evolving extension of the World Wide Web in which information is given well-defined meaning, better enabling computers and people to work in cooperation, making it possible for the web to understand and satisfy the requests of people and machines to use the web content. In the Semantic Web, data itself becomes part of the Web and is able to be processed independently of application, platform, or domain. The vision of the Semantic Web is a web of data that not only harnesses the seemingly endless amount of data on the World Wide Web, but also connects that information with data in relational databases and other non-interoperable information repositories.
Why do we need Semantic Web?
1. Today’s Web
* Natural language
* Information from different sites
* Keywords-based search engine
2. Human can process this easily.It provides a way to connect people and data
* Combine information from diverse sources
* Improve search results by adding context information.By adding additional data, such as metadata, applications can derive more meaningful data from the web
3. However, we want these to be automatically supported by tools
4. The problem is : machines are ignorant!
Central idea:
1. Make the meaning of web content machine accessible and processable.
2. Not to develop super intelligent agents.
3. Attack the problem from the web page side.
The Extensible Markup Language (XML)is a general-purpose specification for creating custom markup languages. It is very flexible text format derived from SGML (ISO 8879). It is classified as an extensible language, because it allows the user to define the mark-up elements. XML's purpose is to aid information systems in sharing structured data, especially via the Internet, to encode documents, and to serialize data; in the last context, it compares with text-based serialization languages.
XML-EXtensible Markup Language
1. A markup language much like HTML
2. Tags not fixed: user definable tags
3. Designed to carry data, not to display data
4. More easily accessible to machines
Limitations of XML
1. No inherent semantics (meaning) of data
2. Up to each application to interpret
3. E.g. tag-nesting interpreted as part-of, or subtype-of, or something else?
RDF—Resource Description Framework
1. A data model to describe resources in WWW
2. Domain independent
3. The fundamental concepts of RDF are:
4. Resources(a “thing”: authors, books, publishers)
5. Properties (describe relations between resources)
6. Statements (an object-attribute-value triple, assert the properties of resources)
RDF Schema(RDFS) is an extensible knowledge representation language, providing basic elements for the description of ontology, otherwise called RDF vocabularies, intended to structure RDF resources. RDFS is used to create vocabularies that describe groups of related RDF resources and the relationships between those resources. An RDFS vocabulary defines the allowable properties that can be assigned to RDF resources within a given domain. RDFS also allows users to create classes of resources that share common properties.
RDF Schema—Vocabulary Description Language
1. Define the vocabulary used in RDF
2. Classes and Properties
3. Classes: courses, lecturers, students, classrooms
4. Properties: teaches, location
5. Class Hierarchies (subClassof)
6. E.g. “staff member”, “professor”
7. Property Hierarchies (subPropertyof)
8. E.g. “involves”, “is taught by”
Limitation of RDFS
1. Limited expressive power
2. Special characteristics of properties (transitive, inverse)
3. Disjointness of classes (male & female)
4. Boolean combinations of classes (union, intersection, complement).
5. Cardinality restrictions (a person has exactly two parents, a course is taught by at least one lecturer)
The OWL Web Ontology Languagenis designed for use by applications that need to process the content of information instead of just presenting information to humans. OWL facilitates greater machine interpretability of Web content than that supported by XML, RDF, and RDF Schema (RDF-S) by providing additional vocabulary along with a formal semantics. OWL has three increasingly-expressive sub-languages: OWL Lite, OWL DL, and OWL Full.
OWL– Web Ontology Language
1. Compatible with RDF Schema
2. More powerful expression ability
3. Three different sublanguages:
4. OWL Full
5. pro: Powerful expressivity, fully compatible with RDF/RDFS
6. con: so powerful as to be undecidable
7. OWL DL
8. pro: permits efficient reasoning support
9. con: restricted expressivity
10. OWL Lite
11. pro: easy to grasp and easy to implement
12. con: restricted expressivity
A semantic network is a graphic notation for representing knowledge in patterns of interconnected nodes and arcs. They are graphical depictions of knowledge of nodes and links that show relationships between object. Computer implementations of semantic networks were first developed for artificial intelligence and machine translation, but earlier versions have long been used in philosophy, psychology, and linguistics.
Common Semantic Networks.
1. Definitional networks emphasize the subtype or is-a relation between a concept type and a newly defined subtype.
2. Assertional networks are designed to assert propositions. Different versions of propositional semantic networks have different syntactic mechanisms.
3. Implicational networks use implication as the primary relation for connecting nodes. They may be used to represent patterns of beliefs, causality, or inferences.
4. Executable networks include some mechanism, such as marker passing or attached procedures, which can perform inferences, pass messages, or search for patterns and associations.
5. Learning networks develop their representations by acquiring knowledge from examples. The new knowledge may change the old network by adding and deleting nodes and arcs or by modifying numerical values, called weights, associated with the nodes and arcs.
6. Hybrid networks combine two or more of the above techniques.
The University of Utah Health Care uses Cerner Millennium(R), PathNet Helix(TM), which incorporates Clinical Bioinformatics Ontology (CBO) a structured vocabulary to describe genetic information, melding the existing medical and research vocabularies so that laboratories may operate with a unified language. The CBO was initiated to address these gaps and covers the areas of molecular genetics, molecular pathology, cytogenetics and infectious disease. The CBO uses a standardized methodology based on consistent application of RefSeq information.
The Reference Sequence(RefSeq) database provides a biologically non-redundant collection of DNA, RNA, and protein sequences. The collection includes sequences from plasmids, organelles, viruses, archaea, bacteria, and eukaryotes. Each RefSeq represents a single, naturally occurring molecule from one organism. Each RefSeq represents a single, naturally occurring molecule from a particular organism.
Cleveland Clinic’s application of semantic web technology
- The Center for Clinical and Translational Science(CCTS) is a Data Coordinating Center.
- CCTS coordinates data collection and reporting of clinical trials data.
- Data<-treatment of diseases with experimental drugs and devices
- Each study’s data is contained within an RDBMS
- Until published, data is inaccessible by clinical data researchers
- This is the current (old) paradigm
Highlights of the University of Utah Center for Clinical and Translational Science: The University of Utah Center for Clinical and Translational Science builds on the University's strengths in genetics and bioinformatics to translate promising bench science into practices that improve human health. The Center will serve as an academic home for clinical and translational research, developing innovative health services for the community and health researchers, providing seed funds to initiate clinical and translational research projects, and training a new generation of clinical and translational investigators.
Goals of the Semantic Web used at the Cleveland Clinic.
- Transform clinical patient data into a well-defined, standardized knowledge representation
- Integrate clinical data for all patient records
- Capture clinical information formally and without ambiguity
- Allow clinical research on abstracted data
- These are closely related to CCTS needs
- The key Semantic Web advantages at the Cleveland Clinic are:
- Use of familiar, local terminology
- Support for unanticipated modeling extensions
- High degree of automation
- High-fidelity integration and mapping with external systems and terminologies
- Support for accurate answering of expressive queries
SemanticDB Processing 1. SemanticDB is interfaced with query engine Cyc Knowldege Base (Cycorp inc.). Cyc is an artificial intelligence project that attempts to assemble a comprehensive ontology and knowledge base of everyday common sense knowledge, with the goal of enabling AI applications to perform human-like reasoning. It is a very large knowledge base that contains over 1.5 Million "facts, rules-of-thumb and heuristics for reasoning about the objects and events of everyday life."
2. Clinical Investigators interacts with CycKB -> builds query fragments. The CYC KB queries use an inference engine that "performs general logical deduction (including modus ponens,modus tolens, and universal and existential quantification)"
3. creates SPARQL queries.
4. run against the OWL patient domain.
5. The exported OWL allows for integration with external systems
Conclusions of the Semantic Web of Cleveland Clinic’s Adoption
1. CCTS captures relatively structured data
2. Cleveland cardiovascular patient data is scattered
3. CCTS, Cleveland both have complex data
4. CCTS stores data in separate silos (relation db), and has the need to be transformed, integrated, have research tools
5. CCTS will migrate towards federated data sharing service similar but larger scope
6. SemanticDB serves as good implementation of Semantic Technologies
7. Demonstrates creating integrated, searchable data
8. Allowing for creation of new knowledge
An Introduction to Handheld Computing in Medicine.
Handheld computers have been used in medicine for more than a decade. Early models like the Apple Newton MessagePad could do many things and showed promise, but their relatively large size, short battery life, and poor handwriting recognition led to a decline in popularity beyond medicine, and they were discontinued. The mid-1990s saw the introduction of other handhelds that have since become the major players. Devices such as the Palm Pilot (launched in 1996) solved many of the problems that plagued earlier models. Being smaller, less expensive, and easier to use and having a significantly longer battery life, these devices were quickly embraced by mobile professionals, including some physicians. Today, medical schools, residency programs, physician groups, and healthcare systems are seeing the potential benefit of these devices for improving health care education and practice and are investing in them.
Some common characteristics of Handheld computing
1. Handheld computing: Pocket-sized computing device, typically having a display screen with touch input a e.g. PDA, Smartphone, Handheld PC
2. Handheld computing in Medicine: Physicians use handheld computers to access reference information, improve coding and billing, and track patient data, all at the point of care.
3. Approximately 60 – 70% of medical students and residents use PDAs for educational purposes or patient care.
Why Physicians Need Handheld Computing in Medicine
1. Medical errors can result from having inadequate access to appropriate information
2. Physicians often find themselves asking clinical questions and making decisions that require ready access.
3. The increased expectations to follow guidelines and formulary restrictions immediately.
What is the Basic Goal of Handheld Computing
It is to provide comprehensive physician-centric solutions that enable clinicians to manage patient information and access medical content at the point of care.
Why is That Goal Important?
1. Through handheld computing, having information readily available at the point of care can be extremely useful in view of the growing amount of medical information available.
2. It can improve efficiency and early data suggest.
3. It can help avoid medical errors and achieve better patient outcomes.
Pattern of PDA Usage
1. Younger physicians in large hospital-based practices are more likely to use PDAs.
2. Use is more related to administrative and organization rather than patient care.
1. Administrative or organization (74%): Billing and coding, calendar scheduling, web and email access, etc.
2. Patient care (61%): drug information, prescribing, accessing patient records, etc.
3. 75% of pediatricians graduating from medical school routinelyuse PDAs.
What are handheld computers good for?
1. Handheld computers hold promise because they are mobile and portable.
2. A host of applications exist for them, the best of which are as portable medical references and medical calculators.
3. Physicians can use handheld computers to improve coding and billing, and track patient data, all at the point of care.
What are they not so good for?
1. Connectivity is limited:wireless networks might do the job, but the infrastructure for providing reliable realtime access via a handheld device is lagging significantly behind the technology for handheld computers overall.
2. Patient care software needs work: although patient care applications are exciting, before purchasing one users should address some practical issues, such as how the program will affect work flow and how it can be integrated into existing computer systems.
3. Security needs to be tightened: users need to assure the privacy, security, and confidentiality of patient data as mandated by the time-honored tradition of medical practice.
Handheld computers currently have various limitations, including
1. small screen size, slow data entry, limited
2. memory, and few security features for protecting sensitive data
Doctors' experience with handheld computers in clinical practice.
Doctors expect handheld computers to become more useful, and most seem interested in leveraging (getting the most value from) their use. Key opportunities with handheld computers included their use as a stepping stone to build doctors' comfort with other information technology and ehealth initiatives and providing point of care support that helps improve patient care.
Additional readings:
1. Health Insurance Portability and Accountability Act From Wikipedia
2. On 02/16/2006, the Department of Health and Human Services issued the Final Rule regarding HIPAA enforcement.
3. A statement about Google Health and HIPAA
4.
Semantic Webs, Semantic networks, and Translational Research. Please refer to:http://semanticwebs.blogspot.com/
Week 16: Grid Computing, PACE and Decision Support
Grid Computing
Grid computing is the latest evolution in providing access to a large set of computational resources. Inspired by the supercomputing community, Grid provides a set of utilities to allow a diverse set of services to work with each other to perform complex tasks.
In the informatics community, GRID technologies are represented by the caGRID project. This project provides access to a large set of cancer data for translational medicine. There are number of services provided from basic gene information to tissue banks, tumor image banks and more. This set of services is constantly evolving as the caGRID project moves forward.
caGRID uses it's own version of Globus, which is a common toolkit for GRID computing. Other grid projects, like Folding@Home, use custom software (screen savers) to perform computation on unused resources. There are other GRID projects for various domains.
However, GRID services require the use of a GRID based toolkit and require a lot of development and validation effort. This limits the usefulness of GRIDs as a general computing resource. A upcoming idea is the notion of cloud computing. Cloud computing allows for a more diverse set of services to be developed and interact by using OS virtualization and Web Service tools.
PACE System
The PACE system is a tool that allows primary care providers to conusel patients on nurtitional information and proper diet and exercise habits. This tool uses a behavior profile gathering method to customize plans for individual patients needs. This is critical, as obesity is a growing problem in the US, with numerous impacts on individual and public health.
Studies have verified the effectiveness of the PACE tool. There is a positive measured impact on diet and activity. The system has good acceptable by patients and clinicans. This tool has a number of advantages, but it relies on self-reporting, which is a limitation on the data.
The PACE system provides a potential model for providing interventions at the point of care. This system depends on the use of IT technology to deliver a informative intervention. This model may be useful for other interventions (e.g. smoking cessation).
Clinical Decision Support Models
Clinical Decision Support is where patient oriented systems are combined with decision making tools to create tools that enable physicans to determine how to care for a patient.
There are a number of decision support models that allow for computers to engage in decision making. These include:
* Model-based patient monitoring (Heldt et al, 2006).
* Bayesian networks and Artificial Neural Networks (Sadja, 2008).
* Fuzzy cognitive maps (Papageogiou et al, 2007)
* Decision trees (Abad-Grau et al, 2008)
* Human factors methods (Saleem et al, 2006)
* Less is more (systems) (Vicente, 2008)
o Ludwig Mies van der Rohe
* Hybrid Sytems (Cohen and Hudson, 2007)
These are all techinques in Machine Learning, in which systems can be trained to detect patterns in data or organize data in a model that allows for decisions to be explored. In the clinical setting, this allows physicians to determine how to use clinical information in the context of patient care. Classic examples include support for choosing appropriate drugs for treatment (antibotics for infections a common case). Other systems have helped with diagnostic support.
Links
GRID Computing
caBIG Home Page This is the home page for most information for caBIG.
JAMIA Article Link to JAMIA article on caGRID.
Folding@Home This is the home for the Stanford Folding project.
Globus Project This is the home page for the Globus toolkit, including links to Grid projects.
caBIG Portal Page This is a portal, showing running caGRID services.
Cloud Computing
Amazon EC2 This is the home page for the Amazon EC2 page.
Google App Engine This is a preview of Google App Engine.
Microsoft Azure This is a link to the new Microsoft Cloud Project.
White Paper on Virtualization This is a simple white paper on OS Virtualization, provided by VMWare, the inventors of modern x86 virtualization.
PACE System
Pace Project This is the home page for the PACE project.
Behavior Risk Factor Surveillance System This CDC project provides information on a number of risk factors in the US. Includes risks for obesity.
References: (Taken from lecture slides)
* Calfas KJ, Long BJ, Sallis JF, Wooten WJ, Pratt M, Patrick K. A controlled trial of physician counseling to promote the adoption of physical activity. Prev Med. 1996;25:225-33.
* Calfas KJ, Sallis JF, Oldenburg B, Ffrench M. Mediators of change in physical activity following an intervention in primary care: PACE. Prev Med. 1997;26:297-304.
* Calfas KJ, Sallis JF, Zabinski MF, Wilfley DE, Rupp J, Prochaska JJ, Thompson S, Pratt M, Patrick K. Preliminary evaluation of a multicomponent program for nutrition and physical activity change in primary care: PACE+ for adults. Prev Med. 2002;34:153-61.
* Farzanfar R. When computers should remain computers: a qualitative look at the humanization of health care technology. Health Informatics J. 2006;12:239-54.
* Patrick K, Calfas KJ, Norman GJ, Zabinski MF, Sallis JF, Rupp J, Covin J, Cella J.Randomized controlled trial of a primary care and home-based intervention for physical activity and nutrition behaviors: PACE+ for adolescents. Arch Pediatr Adolesc Med. 2006;160:128-36.
* Patrick K, Sallis JF, Prochaska JJ, Lydston DD, Calfas KJ, Zabinski MF, Wilfley DE, Saelens BE, Brown DR. A multicomponent program for nutrition and physical activity change in primary care: PACE+ for adolescents. Arch Pediatr Adolesc Med. 2001;155:940-6.
* Prochaska JJ, Zabinski MF, Calfas KJ, Sallis JF, Patrick K. PACE+: interactive communication technology for behavior change in clinical settings. Am J Prev Med. 2000;19:127-31.
* Revere D, Dunbar PJ. Review of computer-generated outpatient health behavior interventions: clinical encounters "in absentia". J Am Med Inform Assoc. 2001;8:62-79.
* Treweek SP, Glenton C, Oxman AD, Penrose A. Computer-generated patient education materials: do they affect professional practice? A systematic review. J Am Med Inform Assoc. 2002;9:346-58
Clinical Decision Support Systems
Machine Learning The classic text in Machine Learning. Dense, CS oriented.
Machine Learning Data Sets A classic set of machine learning data. Used to develop machine learning software.
History of DSS This is a history of decision support systems in business. There is some overlap with clinical decision support.
References (taken from slides)
* Shortliffe, Edward Hance, and James J. Cimino. Biomedical Informatics: Computer Applications in Health Care and Biomedicine. New York, NY: Springer, 2006.
* Kawamoto, CA Houlihan, and EA Balas, et al. Improving clinical practice using clinical decision support systems: a systematic review of trials to identify features critical to success BMJ 2005. Apr 2. 330: (7494) 765.
* Stead, WW; Miller, RA; Musen, MA; Hersh, WR. Integration and beyond: linking information from disparate sources and into work-flow [See comments]. J Am Med Inform Assoc. 2000;7(2):135–45.
* Heldt, T.; Long, B.; Verghese, G.C.; Szolovits, P.; Mark, R.G.; Integrating Data, Models, and Reasoning in Critical Care. Engineering in Medicine and Biology Society, 2006. EMBS '06. 28th Annual International Conference of the IEEE Aug. 30 2006-Sept. 3 2006 Page(s):350 – 353.
* Sajda, P. Machine Learning for detection and diagnosis of disease. Annual Review of Biomedical Engineering. Volume 8, Page 537-565, Aug 2006
* Abad-Grau M.M., Ierache J., Cervino C., Sebastiani P. Evolution and challenges in the design of computational systems for triage assistance (2008) Journal of Biomedical Informatics, 41 (3), pp. 432-441.
* Papageorgiou E. Stylios C, Groumpos P. Novel Architecture for supporting medical decision making of different data types based on Fuzzy Cognitive Map Framework. Engineering in Medicine and Biology Society, 2007. EMBS 2007. 29th Annual International Conference of the IEEE 22-26 Aug. 2007 Page(s):1192 - 1195
* Saleem JJ, Patterson ES, Militello L, Asch SM, Doebbeling BN, Render ML. Using human factors methods to design a new interface for an electronic medical record. AMIA Annu Symp Proc. 2007 Oct 11:640-4.
Drug Decision Support System:
What is Drug Decision Support?
1. It is a computerized assistance to clinicians during physician order entry providing protection against medical errors and lowering medication-related costs.
2. It’s designed to protect against medical errors and to lower medication related costs
3. Computer provider order entry (CPOE) combined with clinical decision support (CDS ).
4. Decision Support can introduced into healthcare organizations at in two stages: Basic and Advanced
1. Different Stages of Decision Support
a. Decision support can be classified into
1. basic:
i. Drug Allergy checking
ii. Basic dosing guidance
iii. Formulary decision support
iv. Duplicate therapy checking
v. Gives alert if patient has allergy that has been documented
vi. Combines coding scheme and medications and allergy data
2. Advanced
i. Advanced dosing support
ii. Guidance for medication-related laboratory testing
iii. Drug-pregnancy checking
iv. Drug-disease contraindication checking
b. Some issues needed to be concerned
a. Low physician acceptance
b. Too many alerts can be overwhelming
Decision Support in Drug Allergy checking
. Current situation
. Presents an alert when a provider orders a medication to which the patient has an electronically documented allergy
i. Input-Coding scheme combining both medications and allergies data
a. Existing problems
. Applications do not distinguish between a drug allergy and a drug sensitivity
i. Excessive drug-allergy alerting in clinically irrelevant circumstances
ii. Poor quality allergy data in the clinical database d. Physicians acceptance - 20%
iii. Override without reason ii. Dosing guidance 60% prescribing errors are wrong dose
b. Suggestions
. Clinicians should contribute to allergy database
i. Coded allergen and coded reaction – clearly discriminating between true allergies and sensitivities/intolerances
ii. Reverse Allergy checking
iii. Avoid over-alerting by improving specificity of alerts and by improving allergy data quality
iv. Asking the clinician to provide a coded override reason whenever he or she overrides drug-allergy alert
v. Analyses of override reasons should occur as part of system quality improvement efforts
Basic Dosing Guidelines
. Clinicians can receive dosing advice by providing a drop list or guidance in written form
a. Leading cause of Adverse Drug Events (ADEs)
0. 60% prescribing errors
b. Mechanisms to improve medication dosing
. Minimal Interference: Offering the clinician a list of patient-appropriate dosing parameters for each specific medication
i. Complete pre-written medication orders that include dose, dose form (when necessary), route of administration, frequency, and a PRN flag and reason.
Formulary Decision Support
. Each institutional drug formulary captures the best clinical judgments of local physicians, pharmacists, and other experts.
a. Include “restricted medications”— drugs that may be used only for a particular indication or require a specialist consult.
b. Mechanisms Followed
. Captures best practices from local experts
i. Include restricted meds that may require a consultant to approve
ii. include in the CPOE order catalog
iii. display a non-formulary or restricted designation with the medication name or within the provided order sentences
iv. display a pop-up alert when the clinician attempts to order a non-formulary drug, while providing a selectable list of alternative formulary medications
Duplicate Therapy Checking
. Drug duplication consists of prescribing more than one regimen of a single drug, or multiple regimens of different medications with similar therapeutic effects
a. Mechanisms
. Duplicate therapy warnings
i. Alerts are in the form of pop-up interruption
b. Problems
. A new inpatient order can trigger a duplicate alert if the order matches a drug on the patient’s ambulatory medication list
i. May not always be clinically relevant
ii. Trigger a duplicate alert for a future order even if the current medication order will expire before the future order begins.
iii. Excessive inappropriate alerting may lead to desensitization to all classes of alerts
iv. Interrupt clinicians workflow and cause frustration when the “duplicate” was already considered and not deemed to be a problem
c. Suggestion:
. Restrict duplicate medication class alerts to those with high risk for adverse events
Drug–Drug Interaction Checking
. Many drugs interact with the metabolism or action of other drugs, causing untoward effects that are best avoided. Thus, checking if drugs affect metabolism or action of other drugs should be an included procedure.
a. Sources, including commercial vendors, supply drug interaction databases to support automated drug–drug interaction checking
b. Problems
. Generate large numbers of clinically insignificant (low severity) alerts that clinicians ignore.
i. When a clinician overrides an alert, it does not uniformly indicate that the alert had no value.
ii. Lack of availability of flexible database to suit hospital requirements
iii. No complete information in warnings
iv. Hard to maintain volume of data
Advanced Clinical Decision Support
Advanced Dosing Guidance in CPOE
1. Required Inputs
1. Basic patient demographics like age, weight, height
2. Indication for the drug
3. Meds that patient is already on
4. Patient’s previous reaction to the current drug
2. No commercially off the shelf systems available now to be integrated with CPOE or as plug-ins
3. Available DSS
1. Antibiotic Assistant (Evans and colleagues)
2. Input : patient’s renal function and age, and cultured organisms’ sensitivity patterns
Advanced Guidance for Medication-associated Laboratory Testing
1. Takes into account important parameters when med is initiated
2. Suggestions for laboratory tests
3. Pre-req for this to work
1. The CPOE application must have access to the patient’s previous laboratory results.
2. CPOE outside the system might be needed to inform providers when a patient is due or overdue for a medication monitoring test.
3. The knowledge bases upon which monitoring recommendations are made should be evidence-based, with documentation of benefits
Advanced checking of drug-disease interactions and contraindications
1. These only work when you have a final analysis for the patient
2. There may be many unknown variables that need to be addressed before this could be successful in assistance
3. Requirements
1. develop an accurate and accountable knowledge base of drug–disease contraindications
2. Contraindication alerts can only work only when patients’ diagnoses and conditions have been finalized and accurately and comprehensively
3. entered as structured data into the EHR
4. Problem
1. Current databases based on contraindication listed in package inserts and these are incomplete and unreliable
Advanced drug pregnancy alerting
1. Identify drugs that are high teratogenic, should never be given to a woman who is, or might be, pregnant
2. Challenges include
1. Pregnancy not routinely performed on admission
2. Many systems don’t even track pregnancy even if test was performed
3. Systems do not update the information when a pregnancy has ended
4. Required to document the stage of the pregnancy
5. Patients having certain conditions cannot become pregnant at any age so considering this option would be inappropriate
Case Study: The impact of academic detailing and a computerized decision support system (CDSS) on prescribing medication at the point of care
Background and Setting
1. Performed at Royal Melbourne hospital and patients with community community acquired pneumonis(CAP) were studied
1. Emergency Department Doctors
2. Focused on antibiotic prescription for patients initially diagnosed with CAP.
3. Focused on Emergency Dept over 3.5 years
4. This study demonstrates the pattern of behavioral change in emergency department clinicians over three and a half years
5. With the rapidly expanding body of medical knowledge, clinicians need access to appropriate, relevant information to guide their clinical decision making
6. A major roadblock is the ability to implement the best system in a busy hospital
Research question
1. What is the impact of different methods of guideline promotion on clinician prescribing behavior?
General Plan of Attack
This study compares how two system impact clinicians' prescribing behavior
Only patients who came through the emergency department were studied, including.
1. Academic Detailing (AD)
2. Computerized Decision Support System(CDDS)
3. Focus of interest
1. Prescription of antibiotics according to guideline recommendations
2. Early identification of severely ill patients
3. Adjustment of antibiotics to meet prescribing recommendation for the severely ill
4. Adjustment of antibiotics to accommodate known patient allergies
Methods
1. These were available but no strict enforcement was used to get cliniicans to adopt these methods ii. Training 1. See slides
Computerized decision support
1. Used to highlight patients with severe pneumonia 2. Hospital db used to rapidly calculate scores for severity of patients
Data collection
1. Inclusion criteria of patients admitted in the ED:
1. Patients with a diagnosis of pneumonia, chest infection, lower respiratory tract infection, pleuritic chest pain, cough, shortness of breath, and/or aspiration
2. Patients who showed a new respiratory symptom, a new chest x-ray consistent with pneumonia, and if the initial assessment made by the treating doctor was that the patient had pneumonia
2. Exclusion criteria
1. Patients under the age of 18
2. Immunocompromised individuals
3. Individuals with SARS
4. Nosocomial pneumonia
5. Suppurative lung diseases (such as cystic fibrosis or bronchiectasis)
3. Information regarding ongoing antibiotic use was collected but not included
4. If date of discharge was not recorded, it was assumed to be five days
5. Antibiotic costs were calculated using pharmacy purchasing data
6. Clinicians were not aware that the study was being conducted
Outcome Measures
1. Primary outcomes
1. Assess the prescription of antibiotic therapy that adequately covered the likely pathogens and was concordant with recommendations
2. Secondary outcomes
1. Patients admitted directly from the ED to the ICU were considered to have the severe form of CAP
2. Patients with allergies were assessed
3. Time between presentation to the ED and the administration of antibiotics was recorded
Results
1. Patients during the CDSS period were older
2. Observed death rate during CDSS period was higher
3. The prescribing behavior over the three time periods was analyzed
4. The average cost of prescribing medicine increased from the first time period to the second, but the cost decreased in the CDSS period
Discussion
1. Demonstrates that the implementation of a computerized decision support system was associated with greater improvement in prescribing practices
2. The use of Academic Detailing did not change the rate significantly
3. Improvement in concordance of prescribing was greatest with the CDS system
4. Better recognition of patients with severe pneumonia
Limitations
1. There was no control group
2. Prescribing patterns make assumptions about practice over similar time periods
3. Allocation of doctors to patients is not structured in the ED
Conclusion
1. The CDSS system more effective than academic detailing
2. Improved antibiotic prescribing practices in a hospital setting due to the use of CDSS
Additional readings:
1. Role of Computerized Physician Order Entry Systems in Facilitating Medication Errors
2. Effects of Computerized Clinical Decision Support Systems on Practitioner Performance and Patient Outcomes
3. Suppurative Diseases of the Lung
4. Improving antibiotic prescribing for adults with community acquired pneumonia: Does a computerised decision support system achieve more than academic detailing alone? – a time series analysis
Personal Health Records, Microsoft and Google
What is a Personal Health Record or PHR?
1. Prescribed meds, allergy info, immunizations, current meds, etc.
2. allergy information;
3. immunizations;
4. diagnoses/problem list;
5. results/ reports;
6. contact / registration information (address, emergency contact, primary care physician, other providers etc)
7. past medical history (surgical, family, social, …)
8. insurance information
Why bother with an e-PHR?
1. Engage patients in health care!
2. Transition from inpatient to outpatient care
What are the types/models of PHR?
1. Stand-alone application
1. Simplest
2. Web sites tether to ePHR-motivation here is economic!
1. Integrated with some app owner organization
2. Only for patients of that institution
3. Challenges of adoption
1. Price
2. Slow adoption of EMR by clinician
3. Websites not tethered to EMR-Web sites that are maintained by third parties which allow patients to enter and access their information
1. Only the patient controls the access to information.
2. Granted by patient to others
Points of concern
1. Privacy, security, and trust not HIPAA covered; PHI will not be used for purposes other than the intended use; federal and state privacy/security regulations)
2. Technical assistance (help, security issues, logoffs, firewalls, passwords etc)
3. Legal issues How will the data be used and can it be trusted clinically?
4. Medical Liability (holding physicians liable for providing care based on incomplete or inaccurate info in PHR, for the act of consulting with patients who lives in a state where physician is not licensed etc)
5. Financial (to determine appropriate future ePHR funding; ability to do e-visit billing for ePHR “encounters”, time-saving clinician practice efficiencies)
Evaluating PHR
1. Privacy
1. Put patient in the driver’s seat to control who sees it
2. Usage and Business Model
2. Interoperability
1. Platform
2. Import / Export with numbers of partners
3. Microsoft allows you to connect to CCD and others and medical devices
4. Microsoft only connects currently to healthcare provider but not EMRs
3. Usability
1. GUI
2. Fitting into the life-styles of the patient
Tour of Microsoft
- 3 clicks and still no data…
- Patient needs to update record with data
- Microsoft allows you to share your records via invitation to others using email and you can put an expiration date on their profiles
Tour of Google
- Google won’t allow you to upload documents
- It connects to affiliated providers to get your information
- Can convert paper records to electronic records
Types of PHR
1.Google Health
2. Microsoft HealthVault
PHR Platforms
One of the major characteristics of a PHR is the platform by which it is delivered, including
- Paper-based PHRs
- PC-Based PHR
- Internet-Based PHR
- Portable-Storage PHR
Afterthoughts
1. PHR to restore open market model?
2. Balance privacy concerns, benefits and usablitiy?
Future of PHR
1. Perhaps this will become more mainstream as patients visit their PCPs, particularly in rural settings where small clinical practices are not tethered to a large, integrated EMR.
Additional Reading
1.Personal health record From Wikipedia
2.Google Health From Wikipedia:
3. Google Health Partner Profiles.
4.Microsoft HealthVault From Wikipedia
5.Personal Health Records: Definitions, Benefits, and Strategies for Overcoming Barriers to Adoption.
6.Overview Personal Health Records
7.The Value of Personal Health Records
8.The Value Of Health Care Information Exchange And Interoperability
Drug Coding Standards
1. Pharmacy information systems are proprietary with unique data structures, etc
1. Theradoc has integrated with 150+ systems/facilities
2. Difference comes from workflows, systems, vendors, versions, formularies
2. Examples of data seen
1. Ibuprofen 800 mg tab PO
2. U of U and Florida hospitals both have Cerner systems implemented at the same time and they both store info completely different
3. The VA hospitals don’t even use the same data types or methodologies
3. As stewards of HIT we need standards for :
1. Interoperability
2. Research
3. Advanced functionality
4. Limitations of non-standardization
1. Not able to group, filter, report
2. Can’t do consistent clinical alerts
3. Research is painful to discern apples to apples comparisons
5. National standards do exist
1. NDC
2. NIH
3. Commercial vendors
1. FDB
2. Etc
6. Standard examples
1. NDC is a mess!
1. 130+ rows for 800 mg Ibuprofen
2. RxNorm
1. Inter-vocabulary glue
2. Public comain (free) and part of the UMLS
3. Current drug info is up to date and accurate, easily searchable
Additional Reading
1.Pharmacy informatics
2.The National Drug Code Directory
3.RxNorm Overview
