Development and validation of a diagnostic model for damp-heat syndrome in upper urinary tract calcium oxalate stones
ZHOU Ying-lin,HU Lei,ZHOU Jian-fu
Abstract:
Objective To develop a machine learning model based on LASSO-Logistic regression to improve the diagnostic accuracy of damp-heat syndrome in patients with upper urinary tract calcium oxalate stones. Methods Clinical data from 343 patients with upper urinary tract calcium oxalate stones were retrospectively collected, and patients were divided into damp-heat syndrome group (n=267) and non-damp-heat syndrome group (n=76) according to traditional Chinese medicine diagnostic criteria. The LASSO-Logistic regression method was used to select key diagnostic features and construct a diagnostic model. The model was internally validated through 1000 bootstrap resamplings and evaluated using the Hosmer-Lemeshow test, receiver operating characteristics curve (ROC curve), and decision curve analysis. Results The LASSO-Logistic regression method identified four features with significant diagnostic value for damp-heat syndrome, including age, recurrence number, serum creatinine, and triglyceride levels. The model demonstrated high discriminative ability (AUC=0.918) and good sensitivity (74.5%) and specificity (92.1%). Conclusion The diagnostic model developed using LASSO-Logistic regression effectively diagnoses damp-heat syndrome in patients with upper urinary tract calcium oxalate stones, providing a novel and reliable diagnostic tool for clinical use. Future research will explore the application and validation of this model in a broader sample and multiple centers.
