Zhao Hong, Li Wanling, Li Junsheng, Li Li, Wang Hang, Guo Jianming
Shanghai Xuhui Central Hospital, Shanghai, China.
Zhongshan Hospital, Fudan University, Shanghai, China.
Front Mol Biosci. 2022 May 4;9:880291. doi: 10.3389/fmolb.2022.880291. eCollection 2022.
The aim of the study was to use machine learning methods (MLMs) to predict the stone-free status after percutaneous nephrolithotomy (PCNL). We compared the performance of this system with Guy's stone score and the S.T.O.N.E score system. Data from 222 patients (90 females, 41%) who underwent PCNL at our center were used. Twenty-six parameters, including individual variables, renal and stone factors, and surgical factors were used as input data for MLMs. We evaluated the efficacy of four different techniques: Lasso-logistic (LL), random forest (RF), support vector machine (SVM), and Naive Bayes. The model performance was evaluated using the area under the curve (AUC) and compared with that of Guy's stone score and the S.T.O.N.E score system. The overall stone-free rate was 50% (111/222). To predict the stone-free status, all receiver operating characteristic curves of the four MLMs were above the curve for Guy's stone score. The AUCs of LL, RF, SVM, and Naive Bayes were 0.879, 0.803, 0.818, and 0.803, respectively. These values were higher than the AUC of Guy's score system, 0.800. The accuracies of the MLMs (0.803% to 0.818%) were also superior to the S.T.O.N.E score system (0.788%). Among the MLMs, Lasso-logistic showed the most favorable AUC. Machine learning methods can predict the stone-free rate with AUCs not inferior to those of Guy's stone score and the S.T.O.N.E score system.
本研究的目的是使用机器学习方法（MLMs）预测经皮肾镜取石术（PCNL）后的无石状态。我们将该系统的性能与盖氏结石评分和S.T.O.N.E评分系统进行了比较。使用了来自我们中心接受PCNL的222例患者（90例女性，41%）的数据。26个参数，包括个体变量、肾脏和结石因素以及手术因素，被用作MLMs的输入数据。我们评估了四种不同技术的疗效：套索逻辑回归（LL）、随机森林（RF）、支持向量机（SVM）和朴素贝叶斯。使用曲线下面积（AUC）评估模型性能，并与盖氏结石评分和S.T.O.N.E评分系统的性能进行比较。总体无石率为50%（111/222）。为了预测无石状态，四种MLMs的所有受试者工作特征曲线均高于盖氏结石评分的曲线。LL、RF、SVM和朴素贝叶斯的AUC分别为0.879、0.803、0.818和0.803。这些值高于盖氏评分系统的AUC（0.800）。MLMs的准确率（0.803%至0.818%）也优于S.T.O.N.E评分系统（0.788%）。在MLMs中，套索逻辑回归显示出最有利的AUC。机器学习方法可以预测无石率，其AUC不低于盖氏结石评分和S.T.O.N.E评分系统。
