Department of Clinical Pharmacy, School of Basic Medicine and Clinical Pharmacy, China Pharmaceutical University, Nanjing, China.
Department of Clinical Pharmacology and Nanjing First Hospital, Nanjing Medical University, Nanjing, China.
J Endourol. 2022 Aug;36(8):1091-1098. doi: 10.1089/end.2021.0783. Epub 2022 Apr 28.
The decision-making of how to treat urinary infection stones was complicated by the difficulty in preoperative diagnosis of these stones. Hence, we developed machine learning (ML) models that can be leveraged to discriminate between infection and noninfection stones in urolithiasis patients before treatment. We enrolled 462 patients with urinary stones and randomly stratified them into training (80%) and testing sets (20%). ML models were constructed using five algorithms (decision tree, random forest classifier [RFC], extreme gradient boosting, categorical boosting, and adaptive boosting) and 15 preoperative variables and were compared with conventional logistic regression (LR) analysis. Performance measurement was the area under the receiver operating characteristic curve (AUC) in the testing set. We also analyzed the importance of 15 features on the prediction of infection stones in each ML model. Sixty-two (13.4%) patients with infection stones were included in the study. On the testing set, all the five ML models demonstrated strong discrimination (AUC: 0.892-0.951). The RFC model was chosen as the final model [AUC: 0.951 (95% confidence interval, CI, 0.934-0.968); sensitivity: 0.906; specificity: 0.924], significantly outperforming the traditional LR model [AUC: 0.873 (95% CI 0.843-0.904)]. Gender, urine white blood cell counts, and urine pH level were the top 3 important features. Our RFC model was the first model for the preoperative identification of infection stones with superior predictive performance. This novel model could be useful for risk assessment and decision support for infection stones.
治疗尿路感染结石的决策因这些结石术前诊断困难而变得复杂。因此，我们开发了机器学习 (ML) 模型，可以在治疗前用于区分尿石症患者的感染性和非感染性结石。我们纳入了 462 名尿路结石患者，并将其随机分层为训练集 (80%) 和测试集 (20%)。使用五种算法（决策树、随机森林分类器 [RFC]、极端梯度提升、分类提升和自适应提升）和 15 个术前变量构建 ML 模型，并与传统的逻辑回归 (LR) 分析进行比较。性能测量是测试集中的接收者操作特征曲线下面积 (AUC)。我们还分析了 15 个特征在每个 ML 模型预测感染性结石中的重要性。研究纳入了 62 名感染性结石患者。在测试集中，所有五种 ML 模型均表现出较强的辨别力（AUC：0.892-0.951）。选择 RFC 模型作为最终模型[AUC：0.951（95%置信区间，CI，0.934-0.968）；敏感性：0.906；特异性：0.924]，明显优于传统的 LR 模型[AUC：0.873（95% CI 0.843-0.904)]。性别、尿白细胞计数和尿 pH 值是前 3 个重要特征。我们的 RFC 模型是第一个用于术前识别感染性结石的模型，具有卓越的预测性能。这种新模型可用于评估感染性结石的风险和提供决策支持。
