Medical expert system with elastic fuzzy logic

We investigate the use of elastic fuzzy logic in a typical medical diagnosis. Fuzzy rules are formulated using common symptoms as antecedents and diagnoses as consequents. Different weighting factors on antecedents are assigned in each rule. This is done to better reflect the fact that in most diagnoses, there are major symptoms among all related symptoms. We also formulated a geometry-mean fuzzification scheme in issuing final decisions. A multiple-pass interactive scheme to retrieve symptom descriptions from patients is used to capture more realistic diagnosis information. An internal medicine expert system with the proposed framework is used to illustrate our design