CT-based radiomics of machine-learning to screen high-risk individuals with kidney stones.

Abstract

Screening high-risk populations is crucial for the prevention and treatment of kidney stones. Here, we employed radiomics to screen high-risk patients for kidney stones. A total of 513 independent kidneys from our hospital between 2020 and 2022 were randomly allocated to training and validation sets at a 7:3 ratio. Radiomic features were extracted using 3Dslicer software. The least absolute shrinkage and selection operator (LASSO) method was used to select radiomic features from the 107 extracted features, and logistic regression, decision tree, AdaBoost, and support vector machine (SVM) models were subsequently used to construct radiomic feature prediction models. Among these, the logistic regression algorithm demonstrated the best predictive performance and stability. The area under the curve (AUC) of the logistic regression model based on radiomic features was 0.858 in the training cohort and 0.806 in the validation cohort. Furthermore, univariate and multivariate logistic regression analyses were performed to identify the independent risk factors for kidney stones, which were gender and body mass index (BMI). Combining these independent risk factors improved the predictive performance of the model, with AUC values of 0.860 in the training cohort and 0.814 in the validation cohort. Clinical decision curve analysis (DCA) indicated that the radiomic model provided clinical benefit when the probability ranged from 0.2 to 1.0. The radiomic model has a good ability to screen high-risk patients with kidney stones, facilitating early intervention in kidney stone cases and improving patient prognosis.