Flower pollination algorithm and multilayer perceptron artificial neural network for heart disease feature selection and classification

Abstract

Heart disease or scientifically known as cardiovascular disease (CVD) is a disease that involves the heart or blood vessels. There are different types of heat diseases and their causes, however the most common one is myocardial infection commonly refers to as heart attack. There are many reasons for heart attack that may be avoidable such as lack of physical fitness and obesity but the unavoidable one is genetic reason. To avoid the serious effect of heart attack and lower the danger of heart failure to patients, early detection of myocardial infection is necessary. Machine learning algorithms such as classification are used in early detection of dieses using historic medical data. Many algorithms are developed for early detection of heart disease, however, because myocardial infection data consists of many features which some of them may not be important to the analysis, there is need to try different alternatives and techniques to come up with the best detection algorithm. In this paper, we proposed flower pollination algorithm and Multilayer perceptron (MLP) Artificial Neural Network (ANN) for feature selection and prediction of myocardial infection. We called this algorithm FPA-ANN. The simulation results of this paper show that FPA-ANN is promising in correct prediction of myocardial infection with 84.2% accuracy.