Machine Learning for Predicting Onset and Progression of Chronic Kidney Disease: A Scoping Review

Chronic kidney disease (CKD) is a major public health problem associated with cardiovascular disease, kidney failure, and mortality. With population aging and rising prevalence of diabetes and hypertension, scalable risk-prediction methods are needed to address the growing CKD burden. We conducted a scoping review assessing machine learning (ML) models published between 2014–2025 predicting incidence and progression of CKD. We identified 28 ML studies predicting CKD incidence (n=6) and progression (n=22). Common algorithms included extreme gradient boosting, random forests, decision trees, logistic regression, and elastic net, with reported area under the curve (AUC) values ranging from 0.75-0.96. Most studies originated from high-income countries and relied on internal validation, while only two performed external validation and advanced toward clinical use. Key limitations included small samples, imbalanced data, limited explainability, and clinical integration challenges. Overall, ML shows strong potential for CKD risk-stratification and prediction, but broader validation and prospective evaluation are needed.