摘要
目的:分析基于深度学习的胎心监护对胎儿窘迫风险的识别。方法:分析2017年1月至2019年12月医院产科收治的1010例孕妇。根据孕妇生产时是否发生胎儿窘迫分为窘迫组和对照组。比较多因素Logistic回归分析模型和深度神经网络模型(DNN)预测孕妇发生胎儿窘迫的预测效能。结果:纳入孕妇有153例发生胎儿窘迫,发生率为16.14%。ROC曲线结果显示多因素Logistic回归模型预测发生胎儿窘迫风险的AUC是0.881,DNN模型预测发生胎儿窘迫风险的AUC是0.974,Z检验结果显示DNN模型的预测效能高于多因素Logistic回归分析模型(P<0.05)。结论:DNN模型下胎心监护对孕妇发生胎儿窘迫有良好预测性,有助于早期识别和处理高危孕妇。
Objective:The identification of fetal distress risk by fetal heart rate monitoring based on deep learning was analyzed.Methods:A total of 1010 pregnant women admitted to our hospital from January 2017 to December 2019 were analyzed.According to whether fetal distress occurred during delivery,the pregnant women were divided into distress group and control group.To compare the predictive efficacy of multivariate Logistic regression analysis model and deep neural network model(DNN)in predicting fetal distress in pregnant women.Results:Fetal distress occurred in 153 pregnant women,with an incidence of 16.14%.ROC curve results showed that the AUC of multivariate Logistic regression model for predicting fetal distress risk was 0.881,the AUC of DNN model for predicting fetal distress risk was 0.974,Z test results showed that the prediction efficiency of DNN model was higher than that of multivariate Logistic regression analysis model(P<0.05).Conclusion:Fetal heart rate monitoring under DNN model has a good predictive ability for fetal distress,which is helpful for early identification and treatment of high-risk pregnant women.
作者
张阔
陈莹
刘丹
ZHANG Kuo;CHEN Ying;LIU Dan(Tangshan Maternal and Child Health Hospital,Tangshan 063000,Hebei Province,P.R.C.)
出处
《中国数字医学》
2021年第10期86-92,共7页
China Digital Medicine
基金
以胎儿窘迫为剖宫产指征的胎心监护图形分析(编号:20191527).
关键词
神经网络模型
胎心监护
孕妇
胎儿窘迫
neural network model
fetal heart rate monitoring
pregnant women
fetal distress
