Medical Diagnosis using Fuzzy Logic

Abstract

This paper presents a generalized and effective methodology for prediction of any disease which has measurable symptoms using fuzzy logic. Medical data has ample of imprecision and ambiguity due to which its difficult to predict the consequences of symptoms at the personal level. We have made an attempt to apply this methodology for diagnosis of heart disease. The dataset was taken from UCI repository. In this approach, all the symptoms that cause a particular disease are fuzzified. For each fuzzy value of a particular symptom, we assign an effect value that denotes the possibility of the occurrence of the disease when the symptom has that fuzzy value. This data is filled in a tabular format which forms the knowledge base for the disease. A knowledge base is built by domain experts who have in-depth knowledge of the subject. When user symptoms are fed to the inference engine, the output is in fuzzy values. A defuzzification module transforms fuzzy output to crisp output. This output denotes the certainty of the presence of disease. A web app prototype is developed for this system with an aim for reaching to general masses.