A Fuzzy Logical Model of Computer-Assisted Medical Diagnosis

Abstract

A model of a computer-assisted diagnostic system using fuzzy subsets has been developed. The physician documents symptom—diagnosis presence relationships and symptom—diagnosis conclusiveness relationships by means of labels of the fuzzy subsets never, almost never, very very seldom, very seldom, seldom, more or less seldom, not known, more or less often, often, very often, very very often, almost always, always. Symptoms are regarded as fuzzy subsets of reference sets. The reference set includes all values the symptom may assume. The degree of membership of a value in the fuzzy subset of a symptom is calculated when the patient’s symptom pattern is available. By means of compositions of fuzzy relations, four different diagnostic indications are determined for every diagnosis under consideration: presence indication, conclusiveness indication, non-presence indication and non-symptom presence indication. By performing the diagnostic process, the system provides the physician with proven diagnoses, excluded diagnoses and diagnostic hints, including reasons for the diagnoses displayed. Proposals for further investigations may also be requested.