Construction of a LASSO regression-based predictive model for recurrence of urinary tract stones

Urinary tract stones are a common urological disease with a high risk of recurrence. This study aimed to develop and validate machine learning models for predicting postoperative stone recurrence.

We retrospectively collected data from patients who underwent surgical treatment for urinary tract stones at the Department of Urology, Xuzhou Central Hospital, Jiangsu Province, from October 2018 to October 2024. Differential variables were first screened, and least absolute shrinkage and selection operator regression was subsequently used to identify key predictors. Six machine learning algorithms, including support vector machines (SVM), random forest (RF), k-nearest neighbors (KNN), eXtreme gradient boosting (XGBoost), LightGBM, and extra randomized trees (ExtraTrees), were used to construct prediction models. Model performance was assessed by receiver operating characteristic (ROC) analysis, calibration analysis, and decision curve analysis (DCA).

A total of 1,000 patients were included in this study for analysis, and 15 variables were ultimately selected to construct the model. Based on the area under curve (AUC) and the DCA results in the test cohort, the XGBoost model demonstrated good performance in this study. The AUC (95% confidence interval) was 0.87 (0.81-0.92). Among all models, XGBoost demonstrated the better calibration (intercept =-0.003, slope =1.045).

The machine learning model developed in this study showed good performance for predicting urinary tract stone recurrence. This model may be useful for postoperative risk stratification and individualized clinical management.