Mathematical Structure of Fuzzy Modeling of Medical Diagnoses by Using Clustering Models

Abstract — An Adaptive-Network-based Fuzzy Inference System ANFIS with different techniques of clustering is successfully developed to solve one of the problems of medical diagnoses, because it has the advantage of powerful modeling ability. In this paper, we propose the generation of an adaptive neuro-Fuzzy Inference System model using different clustering models such as a subtractive fuzzy clustering (SFC) model and a fuzzy c-mean clustering (FCM) model in the Takagi-Sugeno (TS) fuzzy model for selecting the hidden node centers. An experimental result on datasets of medical diagnoses shows the proposed model with two models of clustering (ANFIS-SFC & ANFIS-FCM) while comparing the same model but both with and without clustering models (ANFIS). We obtained better results of average Training error of training and checking data with ANFIS-SFC when we used a Back-propagation model of the Learning Rule, and similarly we obtained the best results with ANFIS-FCM when used with a Hybrid model. Also we have applied SFC & FCM models without ANFIS to get different matrices of cluster centers on medical diagnoses. Finally, we have displayed the surface of MF to each of the ten separate clusters of diseases with values of the objective function