A Clinical Decision Framework Using Fuzzy Rule-Based Classifier for Diabetes Risk Assessment

Diabetes is a potentially life-threatening disease with no existing cure, and whose ongoing spread presents an urgent global problem that this research attempts to help rectify. Valuing the significance of early diabetes detection for the prevention of risking the development of severe complications, the study proposes a comprehensible fuzzy logic-based early diabetes detection model. Two fuzzy classifiers are achieved by combining fuzzy logic and the cosine amplitude method. Follow-up fuzzy rules are derived based on such classifier behavior. A high classification accuracy of 96.47% from testing with a publicly accessible diabetes dataset is shown to be achieved by the developed model, which is superior to many previous approaches. The results confirm that the constructed model is a promising approach for the precise diagnosis of diabetes in clinical practice because the model possesses meaningful prediction power and interpretability.