妊娠期糖尿病发病风险预测模型构建与评价

目的 分析妊娠期糖尿病（GDM）的相关危险因素，建立妊娠期糖尿病发病风险预测模型，为个体GDM发病风险预测及早期干预提供依据。方法 回顾性分析919名患妊娠期糖尿病孕妇与同期分娩的949名未患妊娠期糖尿病孕妇的既往病历资料，采用单因素分析和多因素logistic回归分析孕妇孕早期GDM的独立危险因素，构建GDM发病风险预测模型，采用Homser-Lemeshow拟合优度检验、拟合优度校正图、受试者工作特征曲线 （ROC）对该模型进行评估，采用deLong’s test进行AUC显著性差异检验。结果 基于巨大儿史、糖化血红蛋白、空腹血糖、血红蛋白、白细胞计数、活化部分凝血活酶时间及胚胎移植GDM发病风险模型，拟合程度检验 P =0.443＞0.05，验证集ROC曲线下面积（ AUC ）为0.836（95% CI :0.8019~0.838），灵敏度为78.8%，特异度为78.6%，与训练集ROC曲线下面积无显著性差异（ P ＞0.05）。结论 基于危险因素构建的妊娠期糖尿病风险预测模型有较好的预测性能和泛化能力，能早期预测孕妇GDM发病风险并实施早期干预措施，从而降低GDM发病率。

Analysis of risk factors of gestational diabetes mellitus and establishment of risk prediction model

Objective To analyze the risk factors of gestational diabetes mellitus (GDM), to establish a pregnancy risk prediction model of gestational period, so as to provide a basis for individual GDM risk predictions and early interventions.]Methods The medical records of 919 pregnant women with GDM and 949 pregnant women without GDM were selected and retrospectively analyzed. Single factor analysis and multivariate logistic regression were used to analyze the independent risk factors of GDM in early pregnancy, and the risk prediction model of GDM was established. TheHomser-Lemeshow goodness-of-fit test, goodness-of-fit calibration curve and receiver operating characteristic curve (ROC) were used to evaluate the model. And the significant difference of AUC was tested by Delong’s test.]Results Based on the GDM risk model of birth history of macrosomia, glycated hemoglobin, fasting blood glucose, hemoglobin, white blood cell, activated partial thromboplastin time and embryo transplantation were independent risk factors for GDM (allP<0.05). The established logistic regression model, its fitting optimal testP=0.443 (P>0.05). The area under ROC curve (AUC) of validation set was 0.836 (95%CI: 0.8019-0.838), the sensitivity was 78.8%, and the specificity was 78.6%, which had no significant difference from that of training set (P>0.05).Conclusion GDM risk prediction model based on risk factors has good prediction performance and generalization ability, which can predict the risk of GDM in pregnant women early and then reduce the incidence rate of GDM.