基于机器学习算法构建初产妇合并巨大儿试产结局的预测模型OACSTPCD

Objective:To explore the influencing factors and prediction models of vagi-nal trial of labor outcomes in nulliparous women combined with macrosomia.Methods:To col-lect clinical data of nulliparous women combined with macrosomia performed vaginal trial of la-bor from January 2022 to December 2023 in Anhui Women and Children's Medical Center.Ac-cording to labor outcomes,they were divided into two groups,including vaginal labor group and caesarean section group.Univariate analysis was conducted between two groups,and LASSO a-nalysis further screened the variables.Support vector machine(SVM),general linear model(GLM),K-nearest neighbor(KNN),and random forest(RF)were used to establish predictive models,respectively.The optimal model was ultimately determined.Results:Maternal age,pre-pregnancy and prenatal body mass index(BMI)in the vaginal labor group were lower than those in the cesarean section group.Maternal height,Bishop score,and fetal femur length in the vagi-nal labor group were higher than those in the cesarean section group.The proportion of natural labor in the vaginal labor group was higher than that in the cesarean section group.The differ-ences were statistically significant.The above variables were regressed by LASSO analysis and combined with machine learning algorithms to construct prediction models.Finally,all predic-tion models were comparable by receiver operating characteristic(ROC),and GLM performed the best among the four prediction models,yielding an area under the curve(AUC)of 0.7699.Conclusion:For the population of nulliparous women combined with macrosomia,the prediction model developed using the machine learning method and maternal age,height,pre-pregnancy and prenatal BMI,Bishop score,fetal femur length,and labor mode is able to provide a refer-ence basis for the selection of delivery method for nulliparous women combined with macroso-mia.