Knowledge Engineering in Health Informatics

ISBN-13: 9780387949017 / Angielski / Twarda / 1997 / 265 str.

Homer R. Warner; Dean K. Sorenson; Omar Bouhaddou

Opis

This monograph series is intended to provide medical information scien tists, health care administrators, physicians, nurses, other health care pro viders, and computer science professionals with successful examples and experiences of computer applications in health care settings. Through these computer applications, we attempt to show what is effective and efficient, and hope to provide guidance on the acquisition or design of medical information systems so that costly mistakes can be avoided. Health care provider organizations such as hospitals and clinics are experiencing large demands for clinical information because of a transition from a "fee-for-service" to a "capitation-based" health care economy. This transition changes the way health care services are being paid for. Previ ously, nearly all health care services were paid for by insurance companies after the services were performed. Today, many procedures need to be pre approved and many charges for clinical services must be justified to the insurance plans. Ultimately, in a totally capitated system, the more patient care services are provided per patient, the less profitable the health care provider organization will be. Clearly, the financial risks have shifted from the insurance carriers to the health care provider organizations. For hospitals and clinics to assess these financial risks, management needs to know what services are to be provided and how to reduce them without impacting the quality of care. The balancing act of reducing costs but maintaining health care quality and patient satisfaction requires accurate information about the clinical services."

Spis treści

1 Background and Legacy.- Overview.- Why Build Medical Expert Systems?.- Definitions.- General Design Questions and Related Issues.- State of the Art.- Knowledge Representation and Computation Methodologies.- Rule-Based Medical Expert Systems.- Probabilistic Medical Expert Systems.- Hierarchical Knowledge.- Hybrid Models and Ambitious Adaptations.- 2 The Expert System Model.- to Modeling.- Choosing a System to Model.- Choosing a Model.- 3 Iliad: The Model Used for This Text.- The Frame Concept.- Building Individual Decision Frames.- Bayesian Frames.- Boolean Frames.- Value Frames.- Nested Frames: Clusters.- The Probabilistic Model: Dealing with Uncertainty.- The Bayes Equation.- Ways of Handling the Assumption of Independence.- Probabilistic Information.- Partial Information.- The “Closeness to True/False” Concept.- Information Content.- Passing Information Among Bayesian and Boolean Frames.- Using Partial Information for Decision Making.- Heuristics That Improve the Model.- Risk Flags.- Display Logic.- Data Drivers.- 4 The Data Dictionary: Limiting the Domain of the Model.- Organization of the Dictionary.- Context Versus Concept.- Hierarchical Relationships.- Granularity of the Dictionary.- Modifying the Dictionary.- Knowledge Contained in the Dictionary.- Inferencing from the Hierarchy.- Word Relations.- Data Relations.- 5 The Knowledge Engineering Process.- How to Structure/Model the Knowledge.- The Overall Process.- Knowledge Sources: Advantages and Limitations of Each.- Literature.- Patient Data Repositories.- Expert Opinion.- Which Findings to Include in a Frame.- Probabilistic and Deterministic Logic.- Reasons to Cluster.- Types of Clusters.- Frames That Return a Value.- Estimating Probabilities.- Testing Frames in Isolation.- Sources of Error.- Tools to Facilitate the Knowledge Engineering Process.- Text Editor and Database.- A Working Outline or Hierarchy.- Accessing Normal Values and Frequently Used Numbers.- Accessing the Dictionary.- Maintaining Consistency Between Numerical Estimates.- Relationships Between Frames.- Documenting Sources of Knowledge and the Knowledge Engineering Process.- Saving, Printing, and Statistics.- Combining Frames into a Working System.- Mapping Free Text to a Structured Vocabulary.- Compiling Frames into a Working Knowledge Base.- 6 Evaluation of the Model.- Testing and Refining the Compiled Knowledge.- Appropriateness of Decisions Based on Data Entered by Experts.- Testing with Data Newly Entered from Patient Charts.- Testing with Cases Stored Earlier.- Modifying Source Frames As Required: The Iterative Process.- 7 Applications of the Model.- Modes of Use.- Consultation Mode.- Critiquing Mode.- Simulation Mode.- The User Interface.- Input.- Output.- Browsing Frames.- Viewing and Using the Differential.- Patient Data Window.- Explain Findings.- Most Useful Information.- Minimal Diagnosis.- Bayes Calculator.- Interfaces to Other Knowledge.- Relevant Literature.- Pictures.- Sound.- Animation/Video.- ICD9 Codes.- Other Coding Systems.- Other Expert Systems.- Compromises.- Ease of Data Entry Versus Confusion Regarding “Inferred No”.- Response Time Versus Sophistication of Algorithm.- 8 Lessons Learned.- Teaching Medical Clerks, Physician Assistants, and Other Trainees.- As a Tool for Preauthorization.- As a Screening Tool for Quality Improvement.- Commercial Users of Iliad.- 9 Knowledge Engineering Tools.- Knowledge Acquisition.- Structuring and Coding the Knowledge.- The Dictionary Program.- Frame Authoring.- Syntax Checking and Compilation.- Testing the Knowledge Base.- Summary.- 10 Example Knowledge Bases.- The Knowledge Engineering Class.- Medical and Pediatric HouseCall.- Symptom Analysis.- Deriving HouseCall from Iliad.- Knowledge Engineering for HouseCall.- 11 Future Challenges.- Links to Patient Data: Client Server/Version of Iliad.- Architecture.- Applications.- Benefits.- Future Directions.- References.- Appendices.- 1 Example Hierarchies of Top-Level Diseases (Final Diagnoses) in Various Medical Specialties.- 2 Approximate Estimated Prevalences for Selected Top-Level Diseases in a Family Practice Setting, Categorized by Specialty.- 3 Using the Iliad KE Tool.- 4 Some Example “Domain-Specific” Symptom Lists.- 5 Example Data Relations.- 6 Example Word Relations.