Expertní systém pro řešení problémů laparoskopických a otevřených operací jater

Abstract

This thesis deals with aplication of softcomputing methods in the medical field at the core. It is especially an effort to partial or complete replacement of human thinking by the technique. This concerns in specific case of using expert systems in the liver cancer area and finding dependencies between real and monitored parameters. The thesis use patient data form Faculty Hospital in Ostrava and between those dates are dependences selected by the expert. Expert system was designed with a predetermined structure model Takagi-Sugeno type. Function of this model was verified and confirm as correct. This method is used for a specific group of patient data. It was used methods of nonlinear regression analysis to identify dependencies and search coeficients. The important point was to determine the value of the loos function for each selected dependence and compare real and identified input values. Such dependence which had the criteria of the loos function declared as good enough.