An Expert System Based on Fuzzy Bayesian Network for Heart Disease Diagnosis

Abstract

In this paper, a Bayesian (belief) network with fuzzy probabilities is proposed for heart disease diagnosis. Due to the complexity of relations between the features we used the Bayesian belief network. The fuzzy probabilities are also used because of the multiplicity of initial probability and belonging each of features to their related class. We have used the classification methods for determining the heart diseases class. For depicting the Bayesian network, we applied the K2 algorithm. We comprised the results of our network with the result of the Bayesian network, naive Bayesian, multi-Support vector machine, multilayer perceptron, radial basis function, and k-nearest neighbors. The result showed that our model is more accurate than others.