A Comprehensive Ensemble Approach with Real-Time Performance Metrics for Heart Failure Prediction

Heart failure (HF) remains a significant global health concern, leading to approximately 17 million deaths annually from cardiovascular disease (CVD). Despite advances in predictive models, current approaches often lack the accuracy needed for effective early diagnosis and management. This study utilizes electronic medical records (EMRs) from 5000 patients, each characterized by 13 clinical features, to enhance HF prediction through advanced machine learning (ML) techniques. We propose a novel ensemble learning (EL) approach using a Stacking classifier (SC) with Logistic Regression (LR) as the meta-model, integrating RandomForest (RF), Multi-Layer Perceptron (MLP) and XGBoost (XGB). The ensemble model demonstrates exceptional performance, achieving an accuracy of 99.66% and an F1 score of 99.32%, significantly outperforming individual models. This study highlights the effectiveness of ensemble methods in improving predictive accuracy for HF, providing clinicians with more reliable tools for early diagnosis and intervention. Future research will explore additional model combinations and meta-models to further enhance prediction outcomes.