Multiple Classifiers System for Medical Diagnosis

Data mining helps in decision making. Due to the peculiar feature of the medical profession, physician desperately needs a helping tool to take an efficient and intelligent decision. Good performance, the ability to appropriately deal with missing data and with noisy data (errors in data), the transparency of diagnostic knowledge, the ability to explain decisions, and the ability of the algorithm to reduce the number of tests necessary to obtain reliable diagnosis are the various features desired from the machine learning classifier to solve the medical diagnostic task. Every machine learning method has its own features and no single method can provide all the desired features. We solved this problem by using multiple machine learning methods. In this paper we developed multiple classifiers system which helps the physician in the time of decision making process. Backpropagation algorithm (ANN), K-NN Algorithm (CBR) and Modified towing splitting rule algorithm (CT) are used in this system. We tested the system with three different disease datasets like diabetes, heart disease, breast cancer. It showed better results in reliability and performance which two are most desired features in the medical diagnostic task.