Development and validation of a nomogram to predict impacted ureteral stones via machine learning.

To develop and evaluate a nomogram for predicting impacted ureteral stones using some simple and easily available clinical features.

From June 2019 to July 2022, 480 patients who underwent ureteroscopic lithotripsy (URSL) for ureteral calculi were enrolled in the study. From the eligible study population between June 2019 and December 2020, a training and validation set was randomly generated in a 7:3 ratio. To further evaluate the generalization performance of the nomogram, we performed an additional validation using the data from January 2021 to July 2022. Lasso regression analysis was used to identify the most useful predictive features. Subsequently, a multivariate logistic regression algorithm was applied to select independent predictive features. The predictive performance of the nomogram was assessed using Receiver Operating Characteristic (ROC) curves, calibration curves and decision Curve Analysis (DCA). The Hosmer-Lemeshow Test was utilized to evaluate the overall goodness of fit of the nomogram.

Multivariate logistic regression analysis showed that flank pain, hydronephrosis, stone length/width, HU below (Hounsfield unit density of the ureter center below the stone), HU above/below (HU above divided by HU below) and UWT (ureteral wall thickness) were ascertained as independent predictors of impacted ureteral stones. The nomogram showed outstanding performance within the training dataset, with the area under the curve (AUC) of 0.907. Moreover, the AUC was 0.874 in the validation dataset. The ROC curve, calibration curve, DCA curve and Hosmer-Lemeshow Test suggested that the nomogram maintains excellent clinical applicability and demonstrates commendable performance. Similar results were achieved in the test dataset as well.

We established a nomogram that can be effectively used for preoperative diagnosis of impacted ureteral stones, which is of great significance for the treatment of this disease.

Minerva urology and nephrology. 2024 Aug 02 [Epub ahead of print]

Yuanjiong Qi, Shushuai Yang, Jingxian Li, Haonan Xing, Qiang Su, Siyuan Wang, Yue Chen, Shiyong Qi

Department of Urology, Tianjin Institute of Urology, The Second Hospital of Tianjin Medical University, Tianjin, China., Department of Urology, Tianjin Institute of Urology, The Second Hospital of Tianjin Medical University, Tianjin, China - .