Related Experiment Videos
Comparison of three Knowledge Representation formalisms for encoding the NCEP Cholesterol Guidelines
1Center for Medical Informatics, Columbia University College of Physicians and Surgeons, New York, New York 10032.
Proceedings. Symposium on Computer Applications in Medical Care
|January 1, 1994
Summary
Comparing knowledge representation formalisms for encoding clinical guidelines, this study found CLIPS to be the most intuitive and easiest to use for representing the National Cholesterol Education Panel (NCEP) guideline.
Area of Science:
- Medical Informatics
- Artificial Intelligence
- Knowledge Representation
Background:
- Knowledge Representation (KR) formalisms are utilized for encoding clinical guidelines.
- Direct comparisons of different KR formalisms for this purpose are limited.
Purpose of the Study:
- To compare the suitability of three distinct KR formalisms for encoding clinical care guidelines.
- To evaluate PROLOG, CLASSIC, and CLIPS for guideline representation.
Main Methods:
- The National Cholesterol Education Panel (NCEP) guideline was encoded using three KR formalisms: PROLOG (First Order Logic), CLASSIC (frame-based), and CLIPS (production rule system).
- The accuracy, compactness, debuggability, and ease of use of each representation were assessed.
Main Results:
- All three formalisms (PROLOG, CLASSIC, CLIPS) accurately encoded the NCEP guideline.
- PROLOG offered the most compact representation but was difficult to debug.
- CLASSIC's encoding complexity increased due to the lack of arbitrary disjunction.
- CLIPS provided the most intuitive and user-friendly encoding.
Conclusions:
- CLIPS is recommended as the most suitable KR formalism for encoding clinical guidelines due to its intuitiveness and ease of use.
- While PROLOG is compact and CLASSIC is feasible, their practical application for guideline encoding presents challenges.
- Further research could explore hybrid approaches or extensions to existing formalisms.