A multiclass machine learning framework for chronic kidney disease staging using CTGAN-based synthetic data augmentation and explainable AI

Chronic Kidney Disease (CKD) requires accurate stage-wise prediction for timely intervention, yet most studies focus on binary classification. This study proposes an AI-driven multiclass machine learning framework for CKD staging using estimated glomerular filtration rate (eGFR). A clinically validated UCI dataset was labeled by stage according to National Kidney Foundation guidelines and augmented using CTGAN to address data imbalance and data scarcity. Random Forest, XGBoost, and Multi-Layer Perceptron models were evaluated using 10-fold stratified cross-validation, with Random Forest achieving the highest accuracy of 97.92%. SHAP-based interpretability identified clinically relevant biomarkers, enabling reliable and explainable CKD stage prediction.