A Stacking Ensemble machine learning model achieved approximately 90% classification accuracy for predicting heart failure risk, outperforming a standalone XGBoost baseline.
Does a Stacking Ensemble machine learning model improve binary classification accuracy of heart failure risk compared to a standalone XGBoost classifier?
A Stacking Ensemble machine learning approach achieved approximately 90% accuracy in predicting heart failure risk, demonstrating potential as an automated clinical decision support tool.
Among non-communicable diseases, cardiac disorders account for the highest share of global deaths, claiming roughly 17.9 million lives annually. Identifying at-risk patients before clinical deterioration offers substantial therapeutic benefit, as prompt intervention materially changes prognosis. This study constructs a machine learning pipeline for binary classification of heart failure risk from structured patient data. Two architectures are developed and compared: an XGBoost classifier and a Stacking Ensemble that combines Random Forest and Gradient Boosting as base estimators, with Logistic Regression serving as the second-level meta-learner. Both models are trained and evaluated on the Heart Failure Prediction Dataset, an aggregated repository of 918 records characterised by 11 physiological attributes. The Stacking Ensemble attains approximately 90% classification accuracy, outperforming the standalone baseline. A Flask web interface exposes the trained model for real-time clinical scoring. The results underscore the value of ensemble strategies in building automated decision support tools for cardiac care.
Das et al. (Fri,) conducted a other in Heart failure risk (n=918). Stacking Ensemble machine learning model vs. XGBoost classifier was evaluated on Classification accuracy. A Stacking Ensemble machine learning model achieved approximately 90% classification accuracy for predicting heart failure risk, outperforming a standalone XGBoost baseline.