目的 分析妊娠期糖尿病患者产后糖代谢异常风险预测模型的研究现状，为日后预测模型的构建及验证提供参考与借鉴。 方法 检索PubMed、Web of Science core collection、EBSCO、Ovid、ProQuest、Elsevier ClinicalKey全医学平台、中国知网、万方、中国生物医学文献数据库中有关构建妊娠期糖尿病患者产后糖代谢异常风险预测模型的研究,检索时间为建库至2022年12月。结果 共纳入14个预测模型,模型的预测时长从产后6周至产后15年不等,模型的构建方法包括Logistic回归法，Cox回归法和机器学习。纳入模型均未进行外部验证。 结论 现有妊娠期糖尿病产后糖代谢异常预测模型预测效能较好，但存在预测因子不全面、缺少外部验证等问题。建议未来采用前瞻性队列研究构建预测模型，充分挖掘可干预预测因子，重视模型外部验证，以促进模型在临床中的应用。

Objective To analyze the research on risk prediction models of postpartum abnormal glucose metabolism for patients with gestational diabetes mellitus,so as to provide references for future development and validation of relevant prediction models. Methods The databases of PubMed,Web of Science core collection,EBSCO,Ovid,ProQuest,Elsevier ClinicalKey,CNKI,WangFang and CBM were searched for literature on the development of risk prediction models of postpartum abnormal glucose metabolism for patients with gestational diabetes mellitus.The retrieval time frame was from inception to December,2022. Results A total of 14 prediction models were involved.The prediction length of these models ranged from 6 weeks to 15 years after childbirth.The development methods of these models included Logistic regression,Cox regression and machine learning.None of these models underwent external validation. Conclusions The prediction efficiency of existent prediction models for postpartum abnormal glucose metabolism of gestational diabetes mellitus is good.However,some problems of incomplete predictors and lack of external validation exist.It is suggested that prospective cohort studies should be conducted to develop prediction models,and changeable predictors should be fully explored.External validation of models should be emphasized to promote the application of these models in clinic

R473.71