A Stacking Based Explainable Boosting Model for Early Prediction of Chronic Kidney Disease
C. Kalpana, K. Jagadeesh, A. Vennila, R. Radhika · 2025
Chronic Kidney Disease is a gradual decline in kidney function. A large number of people with chronic renal failure are identified in a more severe phase. It results in the treatment of the patient being postponed, which could be dangerous. This study applies a stackingbased Explainable Boosting Classifier, a novel method for early chronic kidney disease detection that combines the transparency of interpretable machine learning. It is based on generalized additive approaches with pair-wise relationships and the predictive ability of ensemble learning to enable clear feature impact representation. By combining several base learners, the stacking method enhances final prediction accuracy. Classifiers like Logistic Regression, Decision trees, Random Forest, Explainable Boosting classifier, and the proposed Stacking based Explainable Boosting classifier are utilized for assessment. Performance indicators, including Accuracy, Sensitivity, Precision, F1 score and Specificity are used to assess the machine learning models. The Stacking based Explainable Boosting Classifier achieved a perfect score ($\mathbf{1. 0 0 0 0}$) across all metrics, outperforming other models in both predictive performance and interpretability. This study offers a successful means to mitigate chronic kidney disease by helping medical professionals detect disorders promptly.