基于决策树算法构建瘢痕子宫孕妇凶险性前置胎盘预测模型  

A prediction model of dangerous placenta previa in pregnant women with scarred uterus was constructed based on decision tree algorithm

作　　者：曾庆兰 刘青青 赖恩祺 ZENG Qinglan;LIU Qingqing;LAI Enqi(Department of Emergency,Ganzhou Maternal and Child Health Hospital,Jiangxi Province,Ganzhou 341000,China)

机构地区：[1]江西省赣州市妇幼保健院急诊科,江西赣州341000

出　　处：《中国当代医药》2025年第8期8-11,18,共5页China Modern Medicine

基　　金：江西省卫生健康委科技计划项目(202312051)。

摘　　要：目的探讨瘢痕子宫孕妇凶险性前置胎盘(PPP)的影响因素,基于独立危险因素构建决策树(CRT)风险预警模型。方法回顾性选取2023年1月至2024年6月于赣州市妇幼保健院治疗的295例瘢痕子宫孕妇作为研究对象,根据孕妇是否发生凶险性前置胎盘分为PPP组(n=93)和非PPP组(n=202)。通过单因素分析,将有意义的血清指标构建瘢痕子宫孕妇发生PPP的CRT风险预测模型。结果单因素分析显示,年龄、辅助生殖受孕、早产、剖宫产史和胎盘植入与瘢痕子宫孕妇发生PPP有关,差异有统计学意义(P<0.05)。CRT显示,瘢痕子宫孕妇发生PPP预测因子重要性排序排名第一的是年龄,瘢痕子宫孕妇发生PPP的5类风险人群:①剖宫产≥1次、存在胎盘植入,占该节点构成的18.9%;②剖宫产≥1次、无胎盘植入、年龄>34岁,占该节点构成的68.8%;③剖宫产≥1次、无胎盘植入、年龄≤34岁,占该节点构成的18.9%;④无剖宫产史、存在早产情况时,占该节点构成的23.0%;⑤存在早产、存在辅助生殖受孕,占该节点构成的66.7%。ROC曲线的AUC为0.809(95%CI:0.756~0.861),灵敏度为80.60%,特异度为67.30%,最大约登指数为0.479,表明CRT模型的预测效能较好。结论瘢痕子宫孕妇发生PPP受年龄、辅助生殖受孕、早产、剖宫产史和胎盘植入的影响,构建的CRT模型的风险预测性能良好,临床上或可基于此模型对瘢痕子宫孕妇筛查是否发生PPP。Objective To explore the influencing factors of dangerous placenta previa(PPP)in pregnant women with scarred uterus,and build a decision classification regression tree(CRT)risk warning model based on independent risk factors.Methods A total of 295 pregnant women with scarred uterus treated in Ganzhou Maternal and Child Health Hospital from January 2023 to June 2024 were retrospectively selected as the study objects.According to whether the patients had dangerous placenta previa,they were divided into PPP group(n=93)and non-PPP group(n=202).Through univariate analysis,the CRT risk prediction model of PPP in pregnant women with scarred uterus was constructed with meaningful serum indicators.Results Single factor results showed that age,assisted reproduction conception,preterm birth,history of cesarean section and placenta implantation were associated with PPP in pregnant women with scarred uterus,with statistical significances(P<0.05).CRT showed that the ranking of the most important PPP predictors for the occurrence of pregnant women with scarred uterus was age.Five risk groups were selected:①Cesarean section≥1 time,placenta implantation,accounting for 18.9%of the node composition.②Caesarean section≥1 time,no placenta implantation,age>34 years,accounting for 68.8%of the node composition.③Caesarean section≥1 time,no placenta implantation,age≤34 years,accounting for 18.9%of the node composition.④When there was no history of caesarean section and preterm birth,it accounted for 23.0%of the node composition.⑤Preterm birth and assisted reproduction accounted for 66.7%of this node.In the ROC curve,the AUC was 0.809(95%CI:0.756-0.861),the sensitivity was 80.60%,the specificity was 67.30%,and the maximum approximate entry index was 0.479.It shows that the prediction efficiency of CRT model is better.Conclusion The occurrence of PPP in pregnant women with scarred uterus is affected by age,assisted reproductive conception,preterm birth,history of cesarean section and placenta implantation.The prediction per

关 键 词：瘢痕子宫 凶险性前置胎盘 决策树 风险预测 

分 类 号：R714[医药卫生—妇产科学]