冻融胚胎移植结局的列线图预测模型构建与评估

Construction and Evaluation of Nomograph Model for Predicting the Outcome of Frozen Thawed Embryo Transfer

【目的】探讨冻融胚胎移植(FET)结局的列线图预测模型构建与评估。【方法】回归性分析2019年1月至2020年10月在本院诊治且行FET的302例患者为研究对象,共纳入414个FET周期,根据是否妊娠成功分为妊娠组(232个周期)与非妊娠组(182个周期),分析所有患者临床资料,采用多因素Logistic回归模型筛选FET周期妊娠成功的独立影响因素,将独立影响因素纳入R软件建立列线图预测模型,并通过Bootstrap法内部验证与受试者工作特征(ROC)曲线下面积评估模型的区分度与一致性。【结果】414个FET周期中232个临床妊娠,临床妊娠率为56.04%(232/414);采用多因素Logistic回归分析:年龄<30岁、降调人工周期、优质胚胎数3枚为临床妊娠的独立影响因素(P<0.05);基于年龄、子宫内膜准备方案、优质胚胎数建立列线图预测模型,Bootstrap法重复抽样100次的内部验证结果显示校准曲线预测值与实际值基本一致,Hosmer-Lemeshow拟合优度检验=4.842,P=0.774;ROC曲线下面积为0.783(95%CI:0.738~0.827)。【结论】基于FET临床妊娠率的独立影响因素建立的列线图预测模型具有良好的精准度与区分度,可为FET临床妊娠结局的判断提供一定指导价值。

【Objective】To explore the construction and evaluation of nomograph prediction model for the outcome of frozen thawed embryo transfer(FET).【Methods】A total of 302 patients who underwent FET in our hospital from January 2019 to October 2020 were analyzed by regression.A total of 414 FET cycles were included.They were divided into pregnancy group(232 cycles)and non pregnancy group(182 cycles)according to the success of pregnancy.The clinical data of all patients were analyzed.The independent factors affecting the success of FET cycles were screened by multivariate logistic regression model,Independent influencing factors are incorporated into R software to establish nomograph prediction model,and the distinction and consistency between the model and the area evaluation model under ROC curve are internally verified by bootstrap method.【Results】There were 232 clinical pregnancies in 414 FET cycles,and the clinical pregnancy rate was 56.04%(232/414);Multivariate logistic regression analysis showed that age<30 years old,decreased artificial cycles and the number of high-quality embryos were the independent influencing factors of clinical pregnancy(P<0.05);The nomograph prediction model was established based on age,endometrial preparation scheme and the number of high-quality embryos.The internal verification results of bootstrap repeated sampling for 100 times showed that the predicted value of calibration curve was basically consistent with the actual value,Hosmer lemeshow goodness of fit test=4.842,P=0.774;The area under ROC curve was 0.783(95%CI:0.738~0.827).【Conclusion】The nomogram prediction model based on the independent influencing factors of clinical pregnancy rate of FET has good accuracy and differentiation,and can provide certain guiding value for the judgment of clinical pregnancy outcome of FET.