A Comprehensive Analysis of Machine Learning Methods for Predicting Heart Disease

Heart disease is a global issue that significantly impacts populations, highlighting the importance of early detection for individual well-being. Machine Learning (ML) techniques, such as those utilizing the Cleveland Heart Disease dataset, offer cost-effective methods for accurately predicting heart disease. This review focused on sharing the best practices of several ML approaches in the context of patient prognosis, diagnostics, and treatment of heart diseases. It begins with a MinMax Normalized Data Scaling (MMNDS) to preprocess the dataset by addressing noise and missing values. Next, a GeneticBased Crow Search Algorithm (GCSA) identifies an optimal feature subset. Finally, we employ Support Vector Machine (SVM), Naive Bayes (NB), and Random Forest Classifier (RFC) to predict heart disease based on clinical parameters. Furthermore, the proposed classification methods provide excellent results in identifying early heart diseases with performance metrics such as error rate, precision, recall, F1score, and accuracy.