Developmental design of a rule based expert system for diagnosis

Design of a computer aided diagnostic system for thyroid cytopathology diagnosis based on clinical and cytopathological features is presented. The system works on logical reasoning and pattern matching approach. The programming language is C++. It has three components-a knowledge base based on domain rules; patient data incorporation tool and a user interface. The heuristic knowledge database is based on the clinical and cytopathological knowledge of a consultant expert, with the common sense knowledge represented in the form of weights. The program forms diagnosis on the basis of domain rules and the attributed weightage. Explanation to the conclusion forms part of the report. Uncertainity of diagnosis and undiagnosable lesions are separated out. The knowledge base acquisition facility details and the specific program building is elucidated