Prediction of Heart Disease and Improving Classifier Performance Using Particle Swarm Optimization

Abstract

New technologies, like analytics, artificial intelligence, and machine learning have a wide range of effects on industries like healthcare, the automotive, etc. The healthcare industry is one of the most important that demands for more sophisticated methods to correctly diagnose diseases at an earlier stage. These approaches are needed to accurately and early diagnose diseases. Since heart disease is one of the most common ailments in the modern world, early diagnosis is critical for many healthcare professionals to prevent and save patients’ lives. In addition, this strategy enhances the classification performance of heart disease prediction and classification by utilizing particle swarm optimization, or PSO, and four machine learning algorithms. Random forest (RF), logistic regression (LR), support vector machine (SVM), and Naive Bayes (NB) machine learning algorithms are used. Cleveland Dataset is used for performing the analysis with 19 features. In order to increase the performance of the classifiers, the proposed approaches also choose a significant number of features to utilize as data entries. The acquired results demonstrated that the suggested diagnostic method can accurately forecast the risk level of heart disease (accuracy 98%, precision 97%, recalls 96%, and F1-score 97%).