A Hybrid Optimized Model for Predicting and Analyzing Heart Attacks Using Machine Learning

Abstract

In the current world, heart attacks, or myocardial infarctions, are one of the leading causes of deaths, hence early detection and prevention enhance patient outcomes. That is where machine learning (ML) approaches come in: advanced tools for analyzing complex medical datasets and enabling accurate predictions of heart attacks. This paper focuses on the use of several ML algorithms for heart attack prediction in terms of performance, accuracy, and interpretability. It relies on a public dataset that holds information about patients in the form of age, cholesterol level, blood pressure, heart rate, and cardiac history. Data preprocessing is done mainly by dealing with the missing values and feature scaling and selection using correlation analysis as well as recursive feature elimination for selecting the most relevant predictors. The proposed research evaluates five among the best performing ML algorithms along with logistic regression, decision trees, SVM, random forests, and neural networks, using accuracy, precision, recall, F1-score, and ROC-AUC metrics, and highlights the potential of ML in predicting augmentation of heart attack and support clinical decision-making; future work should integrate ML systems into clinical workflows ensuring generalizability on more generalizable populations.