Hybrid Ensemble Learning to Improve Prediction Disease Kidney Chronic

Abstract

Chronic Kidney Disease (CKD) is a major global health issue with a steadily increasing prevalence and high mortality rates. Early detection remains challenging due to non-specific clinical symptoms, often leading to late diagnosis and severe complications such as kidney failure. Machine learning (ML) offers significant opportunities to support early detection and prediction through clinical and laboratory data analysis. However, single models such as Random Forest (RF), Gradient Boosting (GBM), and Support Vector Machine (SVM) still face limitations in generalization and stability when applied to complex and imbalanced datasets. This study proposes a Hybrid Ensemble Learning approach that combines bagging, boosting, and stacking strategies to improve predictive accuracy and robustness. Experimental results using the CKD dataset demonstrate that the Hybrid Stacking model achieves the best performance, with 99% accuracy, 1.0 precision, 0.983 recall, and an AUC-ROC of 0.992. These findings highlight that Hybrid Ensemble Learning, particularly stacking, significantly enhances model sensitivity and reliability, making it a promising tool for supporting clinical decision-making in CKD prediction.