Background/Objective: Chronic kidney disease (CKD) is a major global health concern associated with significant morbidity, mortality, and healthcare burden. This study aimed to develop an explainable machine learning framework that integrates lifestyle, sociodemographic, and biochemical factors for early CKD risk prediction among middle-aged men using public health survey data. Methods: Data from 968 male participants were preprocessed by removing missing values, deriving eGFR and ACR, and labeling CKD status. Five machine learning algorithms, (i.e., Random Forest, AdaBoost, Naïve Bayes, SVM, and XGBoost) were trained and evaluated using accuracy, precision, recall, and F1-score. Model interpretability was assessed using SHAP, LIME, Boruta, and Pearson’s correlation analyses. Results: AdaBoost yielded the best performance (accuracy = 0.7258, F1 = 0.6457, recall = 0.6923), with robust generalization confirmed by the precision–recall curve (AP = 0.715). SHAP and LIME revealed that serum creatinine, blood urea nitrogen, urinary creatinine, and age were major predictors, whereas lifestyle and metabolic indicators such as BMI, sodium and sugar intake, and sleep duration emerged as secondary factors for CKD. Conclusions: This study demonstrates the effectiveness of an explainable machine learning model that integrates lifestyle, sociodemographic and biochemical data for early CKD prediction among middle-aged men. The AdaBoost-based framework shows strong potential for implementation as a clinical decision-support tool within EHR systems and may contribute to personalized and preventive interventions. It emphasizes the growing importance of modifiable behaviors in kidney disease development and supports future work involving multiple cohorts and temporal model expansion to improve risk stratification for individuals at risk of kidney disease.
Valencia et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: