目的 调查孕妇发生先兆流产的影响因素，构建并验证预测模型。方法 回顾性分析2021年1月至2022年12月于兰州市某三级甲等医院产科就诊的204例孕妇的病历资料，根据是否发生先兆流产，将其分为先兆流产组（103例）和对照组（101例）。应用Logistic回归分析孕妇发生先兆流产的影响因素并构建列线图风险预测模型。采用Hosmer-Lemeshow 拟合优度检验、受试者工作特征曲线、曲线下面积及校准曲线来评估模型的拟合优度及预测效能，采用Bootstrap自抽样进行模型内部验证。结果 居住地、低运动量、6~9周孕囊、5~8周人绒毛膜促性腺激素是孕妇发生先兆流产的独立影响因素。对构建的列线图风险预测模型进行Hosmer-Lemeshow检验结果显示，χ²=6.118，P=0.634。4个影响因素的曲线下面积分别如下：低运动量为0.742、居住地为0.707、6~9周孕囊为0.705、5~8 周人绒毛膜促性腺激素为0.757。Bootstrap自抽样内部验证结果提示模型分辨率良好，符合度高。结论 该预测模型预测效果较好，可为先兆流产高危人群的筛查提供参考依据。

Objective To investigate the influencing factors of threatened miscarriage in pregnant women,and to construct and validate a prediction model.Methods Convenient sampling was used to select 204 pregnant women with obstetric treatment at a tertiary A hospital in Lanzhou,Gansu Province from January 2021 to December 2022 as the subjects,who were divided into the threatened miscarriage group (103 cases) and the control group (101 cases) according to the occurrence of threatened miscarriage.Logistic regression was applied to analyze the influencing factors of threatened miscarriage in pregnant women and construct a nomogram risk prediction model.The goodness of fit and predictive performance of the model were evaluated using the Hosmer Lemeshow goodness of fit test,subject working characteristic curve,area under the curve,and calibration curve.Bootstrap sampling was used for internal validation of the model.Results Place of residence,low physical activity,6-9w gestational sac,5-8 w human chorionic gonadotropin were independent influencing factors for threatened miscarriage in pregnant women.The Hosmer Lemeshow test was performed on the constructed nomogram risk prediction model,and the results showed that χ²= 6.118,P=0.634.The areas under the curve of the four influencing factors were as follows:low exercise=0.742,place of residence=0.707,gestational sac for 6-9 w=0.705,and human chorionic gonadotropin for 5-8 w=0.757.The bootstrap sampling internal validation results indicated that the model had good resolution and high compliance.Conclusions With its good predictive effect,the nomogram risk prediction model can provide references for the screening of the high-risk group.

R473.71

甘肃省教育厅教育教学改革项目 (2023gszyjy-025）