An experimental transformation of a large expert knowledge base

Area of Science:

• Medical Informatics
• Artificial Intelligence in Medicine
• Knowledge Representation

Background:

• INTERNIST is a large and comprehensive medical knowledge base for internal medicine diagnosis.
• EXPERT is a general system designed for creating consultation models.
• Transferring knowledge between expert systems is crucial for advancing AI in medicine.

Purpose of the Study:

• To translate a substantial part of the INTERNIST knowledge base into the EXPERT model.
• To evaluate the diagnostic competence of the translated EXPERT model.
• To demonstrate the feasibility of knowledge transfer between large-scale expert systems.

Main Methods:

• Translation of the INTERNIST knowledge base into an EXPERT model.
• Testing the translated model using 431 internal medicine diagnostic cases.
• Analysis of internal representation and reasoning differences.

Main Results:

• The translated EXPERT model demonstrated reasonable competence in final diagnostic classification.
• 431 test cases were used to evaluate the model's performance.
• Despite internal differences, knowledge transfer proved feasible.

Conclusions:

• The experiment successfully demonstrated the feasibility of transferring knowledge between INTERNIST and EXPERT systems.
• The translated model shows practical utility in diagnostic classification.
• This work supports the interoperability of large medical expert systems.