Development of a high-altitude renal disease diagnostic model based on machine learning and multiple biomarker detection: a retrospective study of 19,068 patients.

BACKGROUND AND AIMS: Early detection of chronic kidney disease (CKD) in high-altitude regions remains difficult due to physiological adaptations that may affect conventional biomarkers. This study aimed to develop machine learning-based diagnostic models tailored to high-altitude populations. MATERIALS AND METHODS: This retrospective cohort study included 19,068 individuals living in high-altitude regions (≈ 2300 m) who attended Qinghai Provincial People's Hospital between 2019 and 2022. Regression models were constructed to estimate glomerular filtration rate (GFR) as a continuous outcome, while classification models were developed to diagnose CKD based on KDIGO criteria. Artificial neural networks (ANN), LASSO regression, ridge regression, and linear regression were implemented for GFR prediction. A subgroup of 289 patients undergoing 99mTc-DTPA renal dynamic imaging served as a physiological reference for external model assessment. Classification algorithms included logistic regression, k-nearest neighbors (KNN), support vector machines (SVM), decision trees, naive Bayes, random forest (RF), and ANN. The dataset was split into training (80%) and testing (20%) cohorts using stratified sampling. Five-fold cross-validation was applied for hyperparameter tuning. Model performance was evaluated using AUC, correlation coefficients, precision, recall, and calibration metrics. RESULTS: In regression models predicting GFR, ANN achieved the highest performance (AUC = 0.87), outperforming the CKD-EPI formula. In the 99mTc-DTPA subgroup, ridge regression showed the best discrimination (AUC = 0.88). In classification models, RF demonstrated superior performance (AUC = 0.92). All ML models outperformed traditional GFR equations. CONCLUSION: Machine learning models significantly improve diagnostic accuracy of CKD in high-altitude populations. ANN and RF models demonstrated promising predictive capability, supporting their potential clinical application in high-altitude regions.