中华全科医学2024,Vol.22Issue(4) :651-655.DOI:10.16766/j.cnki.issn.1674-4152.003472

基于超声特征的早期稽留流产预测模型构建分析

Analysis on the construction of early missed abortion prediction model based on ultrasonic characteristics

孔文翠 张瑶佳 刘明松
中华全科医学2024,Vol.22Issue(4) :651-655.DOI:10.16766/j.cnki.issn.1674-4152.003472

基于超声特征的早期稽留流产预测模型构建分析

Analysis on the construction of early missed abortion prediction model based on ultrasonic characteristics

孔文翠 1张瑶佳 1刘明松1
扫码查看

作者信息

  • 1. 湖州市妇幼保健院超声科,浙江湖州 313000
  • 折叠

摘要

目的 超声特征可用于早期稽留流产的预测,但目前尚无研究说明经阴道超声特征预测早期不同孕周稽留流产的价值,因此本研究建立基于超声特征的早期稽留流产的预测模型,并分析其预测效能.方法 纳入湖州市妇幼保健院2021年1月-2022年2月收治的经阴道超声诊断为宫内单胎妊娠孕妇251名为研究对象,通过超声每周测量患者孕囊(GS)、卵黄囊(YS)直径、顶臀长(CRL)和胚胎心率(HR)等超声特征至妊娠10周末,随访并记录妊娠11~12周251例患者稽留流产发生率及GS、YS、CRL、HR水平.采用SPSS 22.0统计学软件中logis-tic 回归数据包筛选早期稽留流产的影响因素,并构建logistic回归预测模型,绘制ROC曲线分析预测效能.结果 根据孕妇是否发生早期稽留流产分为流产组(43例)和继续妊娠组(208例).多因素logistic回归结果显示:临床特征中睾酮(OR=0.452)、雌二醇(OR=1.422)、孕酮(OR=1.442)均为早期稽留流产的预测因子(P<0.05);基于临床特征的预测模型预测早期稽留流产的AUC为0.815,妊娠6周超声特征预测模型为0.801,7周超声特征预测模型为0.926,8周超声特征预测模型为0.906,9周超声特征预测模型为0.883,10周超声特征预测模型为0.924.Delong分析显示,妊娠7周及10周超声特征预测模型的曲线下面积高于基于临床特征的预测模型(P<0.05).结论 基于6~10周超声特征建立的预测模型可用于妊娠11~12周稽留流产的预测,但仅基于妊娠7周及10超声特征构建的预测模型的预测价值高于基于临床特征的预测模型,其余孕周超声特征预测模型预测价值与基于临床特征的预测模型预测价值相当,其中YS是预测早期稽留流产的最强超声特征指标.

Abstract

Objective Ultrasound features can be used to predict early missed abortion,but there is no research to ex-plain the value of transvaginal ultrasound features in predicting early missed abortion in different gestational weeks,so this study established a prediction model of early missed abortion based on ultrasound features and analyzed its prediction effi-ciency.Methods A total of 251 pregnant women diagnosed as intrauterine singleton pregnancy by transvaginal ultra-sound from January 2021 to February 2022 in Huzhou Maternal and Child Health Hospital were included as the research objects.The gestational sac(GS),yolk sac(YS)diameter,crown-rump length(CRL)and fetal heart rate(HR)were detected by the ultrasound until the 10th week of pregnancy.Two hundred and fifty-one patients were followed up and re-corded the incidence of missed abortion in 11-12 weeks and the levels of GS,YS,CRL and HR.Logistic regression data package in SPSS 22.0 software was used to screen the influencing factors of early missed abortion,and a logistic regres-sion prediction model was constructed to draw ROC curve to analyze the prediction efficiency.Results Pregnant women were divided into abortion group(43 cases)and continued pregnancy group(208 cases)according to whether the pa-tients had early missed abortion.The results of multivariate logistic regression showed that testosterone(OR=0.452),estradiol(OR=1.422)and progesterone(OR=1.442)were all predictors of early missed abortion(P<0.05).The pre-diction model based on clinical features predicted the area under the curve of early missed abortion was 0.815,the pre-diction model of 6-week ultrasound characteristics was 0.801,the prediction model of 7-week ultrasound characteristics was 0.926,the prediction model of 8-week ultrasound characteristics was 0.906,the prediction model of 9-week ultra-sound characteristics was 0.883 and the prediction model of 10-week ultrasound characteristics was 0.924.According to Delong analysis,the area under the curve of the ultrasonic feature prediction model at the 7th and 10th weeks of pregnan-cy is higher than that of the prediction model based on clinical features(P<0.05).Conclusion The prediction model based on ultrasound characteristics of 6-10 weeks can be used to predict missed abortion in 11-12 weeks of pregnancy,but the prediction value of the prediction model based on ultrasound characteristics of 7-week and 10-week of pregnancy are higher than that based on clinical characteristics,and the prediction value of other pregnancy ultrasound characteris-tics prediction models is equivalent to that based on clinical characteristics,among which YS is the strongest ultrasound characteristic index to predict early missed abortion.

关键词

早期稽留流产/超声特征/妊娠/预测模型

Key words

Early missed abortion/Ultrasonic characteristics/Pregnancy/Prediction model

引用本文复制引用

基金项目

浙江省医药卫生科技计划(2023KY1183)

出版年

2024
中华全科医学
中华预防医学会,安徽省全科医学会

中华全科医学

CSTPCD
影响因子:1.688
ISSN:1674-4152
参考文献量16
段落导航相关论文