Fuzzy cluster means algorithm for the diagnosis of confusable disease

Abstract

Medical science have been overwhelmed in recent times by uncertainty of one form or the other which have greatly affected the decision making process and as such led to cases of misdiagnosis and in worst cases death. One of such forms of uncertainty is the confusability of symptomatic presentations of diseases due to the fact that they share common symptoms and as such becomes difficult for physician‟s to correctly diagnose them. This difficulty in diagnosis stems from the inability of physicians to quantify the amount of each disease in the confusable disease set depicted by the symptoms. The ultimate goal of medical science is good diagnosis and prevention of diseases and such it is imperative to implement a system to reduce such cases of misdiagnosis which could arise from confusability of disease symptomatic presentations. In this work an expert system driven by the fuzzy cluster means (FCM) algorithms is proposed. The system accepts symptoms as input and provides the degree of membership of each disease in any confusable disease set. Data on alcoholic liver disease were collected and used in the development of the knowledge base. Fuzzy logic and FCM algorithm propelled the inference engine. The system was implemented with CLIPS expert system shell and Java as the front end platform while Microsoft Access was used as the database application. The system gives a measure of each disease within a set of confusable disease. The proposed system had a classification accuracy of 60%.

Keywords: Artificial Intelligence, expert system Fuzzy cluster – means Algorithm, physician, Diagnosis