ファジィ測度を用いた鑑別診断支援システム(一般講演-診断-)

抄録

This paper presents a method of constructing a diagnostic system with fuzzy measure for two-group differentiation. In this method, possibility measure is constructed by using evaluations of each item ,and it is optimized with learning data by genetic algorithm. We have used a Choquet integral in each case and differentiated cases on the basis of each value. To evaluate the performance of our method, we have applied it to differentiation between an early gastric ulcer and an early gastric cancer. As a result, accuracy between 95% and 100% was obtained. Furthermore, optimized possibility measure appropriately reflects doctors' subject.