Comparative Analysis of Machine Learning Algorithms for Heart Attack Prediction

This study aims to develop a robust and efficient prediction model for heart failure detection by using state-of-the-art machine learning and artificial intelligence techniques. The research analyses various patient data that includes demographic and clinical variables to identify patterns and risk factors associated with heart failure. Parameters such as age, gender, cardiovascular indicators, lipid levels, blood glucose and exercise-induced responses were examined to capture the complex interactions that contribute to heart health outcomes. A comprehensive experimental design was used for the study, which included data preprocessing, feature correlation analysis, model training and performance evaluation of a series of machine learning algorithms. The models tested include random forest, gradient boosting (XGBoost), logistic regression, support vector machines (SVM) and neural networks, with a focus on balancing predictive accuracy and computational efficiency. The results show that ensemble methods such as XGBoost and Random Forest provide superior prediction accuracy while being computationally feasible, making them particularly suitable for clinical applications. Neural network architectures, in particular RNN and FNN (post-SHAP), also achieved high accuracy but were associated with significantly higher computational costs. Simpler models such as decision trees and logistic regression were computationally efficient but delivered lower performance metrics than the ensemble methods. These results underscore the potential of XGBoost and Random Forest as optimal models for integrating AI-assisted heart failure diagnostics into real-time medical decision-making processes and offer a compelling balance between precision and practicality in a clinical context.