关键词: 机器学习, 根治性前列腺切除术, 淋巴漏, 疾病预测模型, 危险因素

Abstract:

Objective To identify risk factors associated with lymphatic leakage after laparoscopic radical prostatectomy （LRP） and to develop a machine learning?based nomogram for predicting such outcomes to support clinical prevention strategies. Methods We retrospectively analyzed perioperative data from 248 patients who underwent radical prostatectomy for prostate cancer between January 2020 and January 2024. Independent risk factors were identified through univariate and multivariate logistic regression analyses. A predictive model was developed， and its diagnostic performance was assessed by the area under the receiver operating characteristic curve （AUC）. Five?fold cross?validation was performed to evaluate the model′s generalizability. A nomogram was subsequently constructed to facilitate individualized risk quantification. Results Among the 248 patients， 89 （35.9%） developed lymphatic leakage， while 159 （64.1%） did not. Independent risk factors for lymphatic leakage included intraoperative lymph node dissection （OR = 5.415， 95%CI： 2.167 ~ 13.532， P < 0.001）， intraoperative plasma transfusion （OR = 2.952， 95%CI： 1.524 ~ 5.718， P = 0.001）， and postoperative fasting duration of ≥ 2 days （OR = 1.412， 95%CI： 1.089 ~ 1.829， P = 0.009）. The predictive model showed good discrimination and calibration （AUC = 0.711， 95%CI： 0.647 ~ 0.776， P < 0.001； sensitivity： 0.764； specificity： 0.597）. Model robustness was confirmed through five?fold cross?validation （training set AUC = 0.822； test set AUC = 0.829）. The nomogram provided a clinically useful tool for quantifying individual risk of lymphatic leakage. Conclusions Intraoperative lymph node dissection， plasma transfusion， and postoperative fasting lasting ≥ 2 days are independent risk factors for lymphatic leakage following radical prostatectomy. The validated predictive model demonstrates favorable clinical utility.

Key words: machine learning, radical prostatectomy, lymphatic leakage, predictive model, risk factors