MED INF 406 - Clinical Decision Support Systems Syllabus Instructors: Jerry Lassa MS, Andrew Hamilton MS NP, Erin Kaleba MPH Description: This course provides an introduction to clinical decision support systems in health information technology. The course is divided into three components with the following topics covered: Week 1 1-2 2 Topic 1) Introduction to Decision Analysis - JL • Approaches to decision analysis in health care • Managing uncertainty, probabilistic modeling of decision problems, and predicting future outcomes (Bayes’ Theorem) • Valuing outcomes – individual vs. societal decision-making 3 3-4 4 7 7 7 2) Performance Measurement – EK, AH, JL • Evidenced-based guidelines for clinical practice • Development, endorsement and implementation of performance measures • Technical specifications and EHR-enabled outcomes measurement • Evaluation of EHR impact • Data mining and warehousing • Applied Clinical Informatics (Research) 5 5 5-6 6 6 6 3) Design, implementation and evaluation of CDSS - AH • Active vs. passive decision support • Knowledge vs. non-knowledge based support • HIMSS Implementer’s Guide for deploying CDSS • Technical design issues with expert systems • Classes, codes, nomenclatures and ontologies • Personal Health Records The following three texts will be used in the course: Decision Making in Health and Medicine, Myriam Hunink and Paul Glasziou, 6th printing 2007; Publisher: Cambridge University Press Clinical Decision Support Systems: Theory and Practice, Berner, Eta S. (Ed.), 2nd ed., 2007 Publisher: Springer, Health Informatics Series (springer.com NOT springerpub.com) Improving Outcomes with Clinical Decision Support: An Implementer’s Guide, Osheroff, Pifer, Teich, Sittig, Jenders, 2005; Publisher: Health Information and Management Systems Society (HIMSS) Student Learning Goals: Upon completion of this course, the student will be able to: 1. Describe current uses of medical decision making and decision support systems in health care. 2. Understand the benefits and limitations of medical decision making techniques. 3. Use various decision making and analytic models to solve both structured and unstructured problems. 4. Understand the role of performance measurement in guiding deployment and monitoring impact of CDSS. 5. Understand the basic features, benefits, and limitations of machine learning and intelligent decision support methods in the healthcare environment. 6. Develop a model for a prototype CDSS to address a healthcare problem. Prerequisites: There are no formal prerequisites for the course, but Med Inf 409 Biostatistics and Medical Informatics is highly recommended to be taken before this course. It is assumed that the students have taken their required entry-level courses or the equivalent, especially Med Inf 402 Introduction to Clinical Thinking. The Decision Making in Health and Medicine text provides limited version of TreeAge that students may use for homework assignments and the final project. Students may elect to purchase a full version of TreeAge Pro Suite for advanced decision tree analyses. A discounted student version is available at http://server.treeage.com/treeagepro/purchase/stu.asp. Purchase the limited functionality student license, currently priced at $45. Teaching Method: Asynchronous learning experiences are developed for each weekly session to help students achieve the learning goals. Online, synchronous class sessions will augment the lecture notes and reading materials developed for each session. Learning reflections are provided to assure that students have achieved the learning goals and challenge students to apply what they are learning. Classroom material will be augmented with online discussions, in-class experiential learning exercises, group assignments and projects. The class will be reading-intensive with many technical articles and applications provided, so students are encouraged to set aside sufficient time to read and review materials. Projects: Under the guidance and advice of the instructors, students will identify a CDS application topic of investigation and study around which they will develop a paper worthy of publishing in a peer-reviewed journal. Instructions for the final presentation and paper are available in the Course Information folder of Blackboard. Projects will be evaluated on the content, format, depth of analysis, and evidence of thought which combines recognized best practices with new, creative ideas from the students; as well as the quality of the presentation. Evaluation Method: Students will be evaluated according to the following: class participation, collaboration and contributions to in-class/online discussions (15%), assignments (30%); final project presentation and paper and presentation (35%); and an on-site proctored final exam (20%). Course Calendar Week 1 Decision Making in Healthcare & Managing Uncertainty This session will provide a summary overview of the entire course, establishing the framework from which greater detail will follow in subsequent classes. An overview of decision-making in general, and medical decision-making, in particular will be discussed. Students will be introduced to decision theory, decision analysis and predictive modeling. This session will also cover the more advanced concepts of probability, uncertainty and risk analysis in decisionmaking. Essential probability theory will be reviewed and applications to modeling uncertainty in medical decision-making will be discussed. Learning goals: • Understand approaches to decision making in medicine • Identify the steps in making a decision • Understand types of medical uncertainty: diagnostic, prognostic and treatment and how to estimate uncertainty using rules of probability • Know where to obtain best data on prognosis and prognostic factors Readings required prior to this class: • Decision Making in Health and Medicine: Chapters 1-2 • Articles: o Revolutionary HIT - Cure for Insanity, EHRS Didn't Improve Quality Of OP Care o Decision Analysis in Patient Care • CDSS website bookmarks – review a sampling of these to understand CDSS offerings on the web Assignments: • Exercises 1.2, 2.2 (select only one article on field of interest: diagnosis, prognosis, or therapy), 2.4 • Discussion Board: Post your project preferences (see structured questions online) Week 2 Choosing the Best Treatment using Decision Trees & Probability Revision This session will cover Decision trees and sensitivity analysis as applied to clinical decision making. Probability revision techniques including Bayes’ theorem, sensitivity and specificity, and likelihood ratios will also be reviewed. Clinical decision-making using value judgments and utility theory will be also be briefly covered. Learning goals: • Learn to construct a decision tree for a medical decision, draw conclusions and conduct sensitivity analyses • Learn how to use Bayes' formula to overcome biases in estimating probabilities • Learn how test accuracy (sensitivity, specificity) can help determine posttest probability • Understand how information can be interpreted and used in decision-making when error is present (false-positive and false-negative) • Understand decision-making using value judgments to weigh benefits vs. harms • Learn techniques for valuing outcomes expressed as utility (quantitative measure of personal preference) including Quality-Adjusted Life Years (QALYs), standard reference gamble, time or person trade-off and health indices Readings required prior to this class: • Decision Making in Health and Medicine: Chapter 3-5 • Articles: o Decision Analysis in Critical Care Medicine, Getting Research Findings Into Practice o Interpreting Diagnostic Tests - Odds, Likelihood Ratios and ROC curves o Experimental Test of a Theoretical Foundation for Rating-scale Valuations Assignments: • Exercises 3.1, 4.1, 4.2, 4.4 • Discussion Board: Students should share an article demonstrating use of a decision tree and comment on key variables in the model, model conclusions for a given set of parameters and how it is expected to help a medical provider or patient in decision-making. • Discussion Board: Students should be reviewing one another’s project preferences to identify potential group membership around common project interests; students may reach out to one another to establish a group or may elect to work individually; instructors will help assign remaining students as needed Week 3 Evidence Based Guidelines & Performance Measurement In this session, we introduce the relationship between evidence based guidelines, performance measurement, and clinical decision support. We will explore the national landscape of measurement, including the use of such measures in incentive programs, such as pay for performance. Performance measures developed by national measurement development committees in recent years are shared and linked to clinical decision support in electronic health records. Learning goals: • Understand the relationship between evidence-based medicine and clinical decision support systems • Learn about national efforts to develop, align, and specify performance measures for use in electronic systems • Learn about a typology for categorizing electronic measures (“e-indicators”) of quality and safety Readings required prior to this class: • Articles: o Performance Measures Using Electronic Health Records o The Architecture of Performance Measures – White Paper o e-Iatrogenesis - med errors caused by technology Assignments: • Students should have their final project teams and topics finalized this week. • Discussion board #1: Share an article describing a study of evidence based guidelines and their impact on outcomes of care; describe how adherence to guidelines were measured and whether they were effective. • Discussion board #2: Teams should post an abstract of your proposed final project. Identify team members and describe in approximately 1-2 paragraphs what your project will cover. Week 4 Development, Endorsement and Implementation of Performance Measures We explore several frameworks of performance measurement that have emerged in healthcare information technology, including the use of EHRS to produce measures, and metrics to monitor effectiveness of technology in patient care. Learning goals: • Learn about early inpatient and outpatient attempts to adapt EHRS to produce national measures • Learn about how performance measures are developed in relationship to evidence based practice • Learn about technical specifications for performance measures Readings required prior to class: • Articles: o Performance Measures Using Electronic Health Records Assignments: • Compile info on five clinical and/or system use measures your or a selected organization tracks: • How/why the measure was developed or selected, detail the numerator, denominator, exclusion criteria, current value and goal (internal or external), recent changes in the measure (improvements or declines). Week 5 Design & Implementation of CDS – An Implementer’s Guide An overview of decision support systems in general will be presented, along with examples of clinical decision support systems. Mathematical foundations of decision support will be discussed, including logic and set theory, knowledge based decision-making, searching, and natural language. A practical framework used to design and implement clinical decision support systems will be presented. Learning goals: • Define, compare and contrast information systems, decision support systems and expert systems • Understand basic principles of logic as applied to decision-making • Learn about a framework for applying clinical decision support to improving outcomes in healthcare organizations • Understand various models of decisions support systems, including the logic as well as the basic algorithm employed Readings required prior to this class: • Clinical Decision Support Systems: Ch 1-2 • Improving Outcomes with Clinical Decision Support: An Implementer’s Guide: Introduction and Ch 1-2 • Articles: o 10 Commandments for DSS - JAMIA o Patient Safety and Quality Healthcare articles Assignments: Final Project Pre-Work I: • Use the worksheets in the Improving Outcomes with CDS Guide to help guide the development of your final project. Complete Worksheet 1-1 (Stakeholders, goals, and objectives) on page 17 and Worksheet 2-1 (CIS inventory) on page 37 with your final project group. Complete the worksheets as thoroughly as you can using examples in the Guide to help you. Ask instructors for guidance as needed. Indicate all members of your group at the beginning of the Pre-work assignment, then only one member need submit into Blackboard Assignments. Week 6 CDS Technical Design Issues, Evaluation of Impact & Personal Health Records The lack of use of clinical DSS’s, along with strategies for improving medical decisions with computer-based systems will be discussed. A survey of recent trends in DSS is also presented including PHRs. Learning goals: • Compare and contrast rule-based learning vs. machine based learning • Provide examples of rule-based and machine based decision support systems • Describe the role of artificial intelligence in medicine • Identify the knowledge base, inference engine and interface for various clinical DSS’s • Review data and knowledge representation in decision support systems • Learn about the emerging role of Personal Health Records in HIT Readings required prior to this class: • Clinical Decision Support Systems: Ch 4 • Improving Outcomes with Clinical Decision Support: An Implementer’s Guide: Ch 3-4 • Articles: o HIT for Improving Quality of Care in Primary Care Settings o The Impact of Electronic Health Records on Time Efficiency of Physicians and Nurses: A Systematic Review Assignments: Final Project Pre-Work II: • Use the worksheets in the Improving Outcomes with CDS Guide to continue development of your final project. Complete Worksheet 3-1 (CDS intervention selection and workflow opportunity) on page 67 and Worksheet 4-1 (CDS intervention specification) on page 79-80 with your final project group. Indicate all members of your group at the beginning of the Pre-work assignment, then only one member need submit into Blackboard Assignments. Week 7 Evaluation of Impact, Data Mining/Warehousing & Applied Clinical Informatics Approaches for evaluating impact of EHR and CDS will be discussed. In addition, a practical overview of data mining applications used for CDSS is presented and the potential for leveraging an EHRS to support research efforts is introduced. Learning goals: • • • • Learn about various models for evaluating the impact of CDSS Explore the role of data mining in developing and refining decision support systems Understand the opportunities and barriers for using an EHRS to support research Critique the impact of EHRs on medical practice, quality of care, and user and patient satisfaction Readings required prior to this class: • Clinical Decision Support Systems: Ch 3, 7 and 11 • Articles: o A systematic review of computer-based patient record systems and quality of care: more randomized clinical trials or a broader approach? o The Effects of Promoting Patient Access to Medical Records: A Review o Overview of Data Warehousing and OLAP Assignments: • Online discussion TBD. • Student Presentations for final project (as needed prior to next week). Week 8 * Final Project Presentations * Week 9 Final exam week.