Heart-disease Prediction Using Ensemble-Learning Techniques

Abstract

Heart-disease poses a growing threat to global health, driven by various lifestyle factors and demographic shifts. This study proposes ensemble model approach utilizing three Kaggle-sourced datasets to anticipate cardiac problems. Our methodology combines Random-Forest, XGBoost, and Logistic-Regression algorithms, alongside a Voting Classifier, to enhance predictive accuracy. Through 5-fold cross-validation, our ensemble model achieves a compelling training accuracy of 99.4% and testing accuracy of 91.7%. This amalgamation of datasets offers a thorough comprehension of multifaceted risk factors, including lifestyle behaviors, genetic predispositions, and clinical markers. Using machine-learning algorithms, our approach empowers healthcare practitioners with actionable insights for early detection and tailored intervention strategies. As heart-disease prevalence continues to rise, integrating advanced ensemble techniques holds promise in improving risk assessment, thereby mitigating its impact on public health.