Refinement of the HEPAR expert system: tools and techniques

Area of Science:

• Software Engineering
• Artificial Intelligence
• Medical Informatics

Background:

• Traditional software engineering has established methods for static and dynamic verification and validation.
• Expert systems, particularly rule-based ones, often lack comparable development support tools.
• Ensuring expert systems meet specifications is more challenging than traditional software.

Purpose of the Study:

• To address the need for better development support in expert systems.
• To illustrate the importance of methods and tools through the HEPAR system development.
• To present a novel approach for the verification and validation of rule-based expert systems.

Main Methods:

• An incremental development methodology was adopted for the HEPAR system.
• Dynamic validation was performed after implementing parts of the expert system.
• Software tools were developed as extensions to a rule-based expert-system shell.

Main Results:

• The implemented tools provide insights into knowledge base modifications.
• These tools identify areas within the knowledge base that require refinement.
• The HEPAR system, a rule-based expert system for liver and biliary tract diagnosis, benefited from these tools.

Conclusions:

• The developed software tools are valuable for refining expert systems.
• Similar tools can aid in the development of other expert systems.
• Enhanced support through specialized tools is crucial for robust expert system development.