Improving Heart Disease Prediction Using Feature Selection Through Genetic Algorithm

Abstract

AbstractHeart disease is one of the leading causes of fatality. A reliable and robust prediction system is needed for people to take preventive measures and medication beforehand and develop a proactive lifestyle accordingly. Various vital features determine human heart health, and it is important to recognize the critical ones that could be determining the chances of getting heart disease in the future. The various machine learning algorithms based on the critical features could predict heart disease more accurately. This article employs evolutionary algorithms like Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) for the feature selection to improve the accuracy of machine learning algorithms further. GA and PSO combined with Naïve Bayes (NB), Support Vector Machine (SVM), and J48 have been applied for feature selection. After selecting the significant features, the effectiveness of the feature selection algorithm is evaluated by applying machine learning approaches on the complete dataset and reduced dataset. Five different machine learning approaches, viz., NB, SVM, Decision Tree (DT), Logistic Regression (LR), and Random Forest (RF) algorithm, have been used to predict heart disease and thus measure the effectiveness of the feature selection approaches. The results indicate that the GA has been the most effective algorithm for feature selection as it enhances the prediction accuracy most.KeywordsFeature selectionGenetic Algorithm (GA)Heart disease predictionClassificationParticle Swarm Optimization (PSO)