武警医学2025，Vol.36Issue(3)：232-238,7.
妊娠糖尿病风险列线图预测模型的构建与验证
Construction and verification of a risk nomogram prediction model for gestational diabe-tes mellitus
摘要
Abstract
Objective To screen the related risk factors of gestational diabetes mellitus(GDM),construct a prediction model of the risk of GDM in pregnant women,and verify the predictive efficacy of the model.Methods The clinical data of 891 pregnant women who delivered in Beijing Obstetrics and Gynecology Hospital from January 2017 to January 2019 were retrospectively analyzed.In order to establish and verify the prediction model,the research objects were divided into two subsets,namely modeling set(623 ca-ses)and validation set(268 cases),by random number method and 7:3 ratio.According to the results of 75g glucose tolerance test during 24-28 weeks of gestation,pregnant women were divided into GDM group and normal glucose tolerance group.The age,fasting blood glucose(FBG),white blood cell count(WBC),neutrophil count(NE),lymphocyte count(LY),neutrophil lymphocyte ratio(NLR),hematocrit(HCT),serum ferritin(Fe),D-dimer(D-D),fibrinogen(FIB),total cholesterol(Cho),triglyceride(TG),high-density lipoprotein(HDL)and low-density lipoprotein(LDL)were compared between the two groups.LASSO regression was used to optimize the screening of variables,and multi-factor logistic regression analysis was used to establish the prediction model and draw a nomogram.Receiver operating characteristic(ROC)curve,calibration curve and Hosmer-Lemeshow goodness of fit test were used to verify and evaluate the discrimination and calibration of the prediction model,and decision curve analysis(DCA)was used to evaluate the clinical validity of the prediction model.Results LASSO regression analysis was used to screen out 12 predictive variables:age,NE,Ly,FBG,HCT,Fe,D-D,FIB,Cho,TG,HDL and LDL.Multivariate logistic regression analysis was performed on these 12 predictors and the results showed that age[OR=1.123,95%CI:1.058-1.192],NE[OR=1.216,95%CI:1.061-1.393],FBG[OR=5.528,95%CI:2.907-10.83],TG[OR=1.193,95%CI:0.949-1.469]and HDL[OR=0.19,95%CI:0.069-0.496]were risk factors for GDM(P<0.05).Based on these 5 risk factors,a nomogram was drawn,and a prediction model was constructed.The area under ROC curve(AUC)for the prediction of GDM in the modeling set was 0.772[95%CI(0.7211,0.8235)],and 0.747[95%CI(0.6597,0.8352)]for GDM in the validation set,with no significant difference in AUC between modeling set and validation set(P=0.6325).Hosmer-lemeshow goodness-of-fit test showed a good fit(modeling set P=0.8335;validation set P=0.07015).When the DCA threshold probabilities were between 0.01-0.55 and 0.01-0.50,the model set and the verifica-tion set had clinical practical value,and the prediction model was beneficial in clinic.Conclusions A nomogram prediction model with five predictors(Age,NE,FBG,TG,HDL)can be used to predict the risk of GDM in pregnant women.关键词
妊娠期糖尿病/危险因素/列线图/预测模型Key words
gestational diabetes mellitus/risk factors/nomogram/prediction model分类
医药卫生引用本文复制引用
兰雪丽,姜海利,崔建芬..妊娠糖尿病风险列线图预测模型的构建与验证[J].武警医学,2025,36(3):232-238,7.基金项目
首都医科大学附属北京妇产医院中青年学科骨干培养专项(FCYY202002) （FCYY202002）
