Analysis of Factors Related to the Prediction of Renal Partial Nephrectomy Clamping Time Based on a Single Hidden Layer Neural Network and R.E.N.A.L.Score

CHEN Huajin; LV Dongshan; ZHOU Lipo; GUAN Youliang; WANG Chengcai; SHENG Yuwen; TONG Ming; JIN Yanyang

doi:10.3969/j.issn.2096-305X.2025.03.016

Abstract：Objective To construct and evaluate a prediction model for renal partial nephrectomy clamping time based on a single hidden layer neural network,and to explore the influencing factors,providing a basis for preoperative precision assessment.Methods This study conducted a retrospective analysis of the clinical data of 328 patients with localized renal tumors who underwent laparoscopic partial nephrectomy at the Department of Urology in the First Affiliated Hospital of Jinzhou Medical University from December 2018 to December 2023.Patients were divided into two groups based on intraoperative renal artery clamping time:the<20-minute group(145 cases)and the>20-minute group(183 cases).Variables included demographic characteristics,preoperative imaging parameters,and perioperative indicators.A neural network model was built using the R language and the tidymodels framework.Cross-validation was employed to optimize the hyperparameters and the model performance was evaluated using AUC,accuracy,and ROC curve analysis.The SHAP(Shapley Additive Explanations)values and partial dependence plots were used to analyze and explain the impact of key variables on prediction outcomes.Results The neural network model achieved an AUC of 0.999 in the training set and 0.883 in the test set,indicating high accuracy and good generalization ability.The importance analysis revealed that tumor size was the primary factor influencing renal artery clamping time,followed by the Nscore collection systerm and longScore(longitudinal position score).There were no significant differences in demographic characteristics,such as gender,age,and BMI,between the two groups(P>0.05).However,significant differences were observed in tumor size,Nscore and longScore(P<0.05).Conclusion The renal partial nephrectomy clamping time prediction model based on a single hidden layer neural network exhibits high accuracy and reliability,offering strong support for preoperative individualized decision-making.Tumor size and relevant imaging indicators significantly influence clamping time and should be closely monitored in clinical practice.