妊娠期糖尿病早期可视化风险预测模型的构建与效能验证  

Construction and Efficacy Verification of an Early Visual Risk Prediction Model for Gestational Diabetes Mellitus

作　　者：沈谦 沈冰燕 SHEN Qian;SHEN Bingyan(Department of Gynecology and Obstetrics,Qidong City Maternal and Child Health Hospital,Nantong,Jiangsu 226200 China)

机构地区：[1]启东市妇幼保健院妇产科,江苏南通226200

出　　处：《中国医药指南》2025年第1期10-13,共4页Guide of China Medicine

基　　金：2023年启东市农业和社会民生科技计划项目(202330)。

摘　　要：目的构建早期可视化预测妊娠期糖尿病(GDM)的风险模型,并验证其预测效能。方法采用随机数字表法,从2022年10月至2023年10月启东市妇幼保健院妇产科收治的孕产妇中选取400例作为观察对象,根据随访结果是否发生GDM分为GDM组(n=89)和非GDM组(n=311),收集并分析两组临床资料、血常规、肾功能及75 g口服葡萄糖耐量试验(OGTT)指标的差异性,采用二元Logistic回归分析GMD独立危险因素,采用R3.5.0软件构建孕产妇GDM风险列线图模型,采用ROC曲线和Hosmer-Lemeshow对列线图模型进行效能验证。结果GDM组与非GDM组身高、饮酒、高血压家族史、职业、红细胞计数(RBC)、血小板计数(PLT)比较(P>0.05),GDM组年龄大于非GDM组,GDM组孕前BMI≥25 kg/m2、被动吸烟、糖尿病家族史所占比例高于非GDM组,GDM组孕次多于非GDM组,妊娠前生殖道炎症和激素用药史所占比例高于非GDM组,GDM组白细胞计数(WBC)、血红蛋白(Hb)、尿酸(UA)和肌酐(Cr)水平高于非GDM组(P<0.05);二元Logistic回归分析结果显示,孕前BMI≥25 kg/m2、妊娠前生殖道炎症、糖尿病家族史、激素用药史、WBC、Hb、UA及Cr是GDM的独立危险因素(P<0.05)。将上述结果纳入构建列线图预测模型,ROC曲线下,在训练集上,AUC值为0.817;在验证集上,AUC值伪0.802。结论早期可视化预测GDM的风险模型预测效能较好,能丰富GDM影响因素研究,并为临床干预GDM提供参考依据。Objective To construct an early visual risk model for gestational diabetes mellitus(GDM)and verify its predictive efficacy.Methods Random number table method was used to select 400 pregnant women admitted to the obstetrics and Gynecology department of Qidong Maternal and Child Health Hospital from October 2022 to October 2023 as observation objects.According to the follow-up results,they were divided into GDM group(n=89)and non-GDM group(n=311),clinical data,blood routine,renal function and 75 g oral glucose tolerance test(OGTT)were collected and analyzed.The independent risk factors of GMD were analyzed by binary Logistic regression.The risk nomogram model of GDM in pregnant women was constructed by R3.5.0 software.The effectiveness of the nomogram model was verified by ROC curve and Hosmer-Lemeshow.Results Height,alcohol consumption,family history of hypertension,occupation,red blood cell count(RBC)and platelet count(PLT)were compared between the GDM group and the non-GDM group(P>0.05).The age of the GDM group was higher than that of the non-GDM group.The proportion of pre-pregnancy BMI≥25 kg/m2,passive smoking and family history of diabetes in the GDM group was higher than that of the non-GDM group.The number of pregnancies in GDM group was more than that in non-GDM group,the proportion of pre-pregnancy genital tract inflammation and hormone use history was higher than that in non-GDM group,and the levels of white blood cell count(WBC),hemoglobin(Hb),uric acid(UA)and creatinine(Cr)in GDM group were higher than those in non-GDM group(P<0.05).Binary Logistic regression analysis showed that pre-pregnancy BMI≥25 kg/m2,pre-pregnancy genital tract inflammation,family history of diabetes,history of hormone use,WBC,Hb,UA and Cr were independent risk factors for GDM(P<0.05).The above results were incorporated into the construction of the nomogram prediction model.Under the ROC curve,the AUC value on the training set was 0.817.On the validation set,the AUC value is pseudo 0.802.Conclusions The early visual risk mode

关 键 词：妊娠期糖尿病 列线图 危险因素 风险预测模型 

分 类 号：R714.256[医药卫生—妇产科学]