基于决策树算法的高龄产妇产后早期盆底肌力损伤预测模型构建OA
Construction of a predictive model for early postpartum pelvic floor muscle weakness in advanced maternal age using decision tree algorithms
目的 分析高龄产妇产后早期盆底肌力损伤的影响因素,并构建决策树模型.方法 回顾性选取 2022 年 3 月至 2024 年 6 月萍乡矿业集团有限责任公司总医院进行 6～8 周产后复查的高龄产妇为研究对象,根据测定结果将其分为损伤组和非损伤组,采用单因素分析和二元 logistic 回归分析损伤因素,构建决策树模型,分析模型效能.结果 有 226 名(56.50%)发生盆底肌力损伤.单因素结果显示,分娩次数、第二产程时间、分娩方式、是否进行孕晚期运动训练、是否进行盆底康复训练、新生儿体重与肌力损伤相关;二元 logistic 回归模型进行多因素分析结果显示,分娩次数≥2 次(β=0.480,OR=1.615,95%CI:1.052～2.479)、第 2 产程时间≥2 h(β=0.774,OR=2.169,95%CI:1.425～3.300)、新生儿体重≥4 kg(β=0.463,OR=1.589,95%CI:1.042～2.422)是高龄产妇产后早期盆底肌力损伤危险因素,进行孕晚期运动训练(β=-0.589,OR=0.555,95%CI:0.362～0.850)、盆底康复训练(β=-0.478,OR=0.620,95%CI:0.407～0.944)是高龄产妇产后早期盆底肌力损伤保护性因素.构建决策树模型,共 5 层,11 个节点,模型选择 6 个临床特征作为节点,重要性排名第一是第 2 产程时间.决策树模型 AUC 为 0.660(95%CI:0.607～0.713),二元 logistic 回归模型的 AUC 值 0.674(95%CI:0.621～0.727).结论 本研究构建的高龄产妇产后早期盆底肌力损伤的决策树模型预测效能较为优异,有利于早期精准鉴别高龄产妇产后早期盆底肌力损伤的发生.
Objective To analyse factors influencing early postpartum pelvic floor muscle weakness in advanced maternal age women and construct a decision tree model.Methods Retrospectively,high-aged pregnant women who underwent 6-8 weeks postpartum follow-up at Pingxiang Mining Group Co.,Ltd.General Hospital from March 2022 to June 2024 were selected as the research subjects.Based on the measurement results,they were divided into the injury group and the non-injury group.Univariate analysis and binary logistic regression analysis were used to identify the injury factors,and a decision tree model was constructed,the model's efficacy was analyzed.Results A total of 226 parturients(56.50%)had pelvic floor muscle strength injury.Univariate analysis showed that the number of deliveries,second stage of labor duration,delivery mode,whether to perform late pregnancy exercise training,whether to receive pelvic floor rehabilitation training,and neonatal weight were associated with muscle strength injury.The results of multivariate analysis by the binary Logistic regression model showed that the number of deliveries was≥2(β=0.480,OR=1.615,95%CI:1.052-2.479),and the dura-tion of the second stage of labor was≥2 hours(β=0.774,OR=2.169,95%CI:1.425-3.300),neonatal weight≥4kg(β=0.463,OR=1.589,95%CI:1.042-2.422)are risk factors for pelvic floor muscle strength injury in the early postpartum period of elderly parturients.Exercise training in the third trimester of pregnancy(β=-0.589,OR=0.555,95%CI:0.362-0.850)and pelvic floor rehabilitation training(β=-0.478,OR=0.620,95%CI:0.407-0.944)are protective factors for pelvic floor muscle strength injury in the early postpartum period of elderly parturients.A decision tree model was constructed,consisting of 5 layers and 11 nodes.The model selected 6 clinical features as nodes,and the time of the second stage of labor ranked first in importance.The AUC of the decision tree model was 0.660(95%CI:0.607-0.713),and the AUC of the binary Logistic regression model was 0.674(95%CI:0.621-0.727).Conclusion The decision tree model constructed in this study for predicting early postpartum pelvic floor muscle weakness in older mothers demonstrated favourable predictive performance,facilitating early and precise identification of such weakness in this population.
张海霞;黄海莲;文娜
萍乡矿业集团有限责任公司总医院中医肾内科,江西 萍乡 337000萍乡矿业集团有限责任公司总医院中医肾内科,江西 萍乡 337000萍乡矿业集团有限责任公司总医院中医肾内科,江西 萍乡 337000
医药卫生
高龄产妇盆底肌力损伤影响因素决策树算法
Advanced maternal agePelvic floor muscle weaknessInfluencing factorsDecision tree algorithm
《中国当代医药》 2026 (10)
29-33,5
江西省卫生健康委科技计划项目(20204471).
评论