DefinePK

Abstract

The medical expert system is a special type of recommender systems that plays a major role in decision making process by the medical doctors nowadays. This kind of expert systems often provides the medical diagnosis activity based on the handling various patients in various situations by the medical doctors and clinical symptoms of patients to give a list of possible diseases attended with the membership values. Many acquiring diseases from that list are then determined by medical doctors experience expressed through specific combinations of features in the clinical dataset. The major issue of the expert system is increasing the accuracy of the medical diagnosis attributes that involves the cooperation of decision making systems and recommender systems in the sense that predict the behaviors of disease symptoms and the doctors experience are represented by rules whilst the prediction of the possible diseases is identified by the prediction capability of medical expert systems. From the past observation, the accuracy of features similarity could be improved by the integration with the information of possibility of patients belonging to clusters specified by a weighted K-means clustering method. For improving the performance of medical expert system, a new hybrid intrusion detection framework is introduced to improve the classification accuracy. This hybrid system is combining the proposed genetic based data adaptation decision tree (GDADT) and the existing weighted K-Means clustering. Moreover, we have used the existing cluster and decision tree based classifier called Intelligent Agent based Enhanced Multiclass Support Vector Machine (IAEMSVM) for improving the prediction accuracy. The experimental results of the proposed system show that this system achieved high-detection rate with less time and low false alarm rate when tested with UCI Machine Learning data set.