基于神经网络的早期先兆流产妊娠结局预测  

Pregnancy outcome prediction of early threatened abortion based on neural network

作　　者：吴睿嘉 陈海燕[2] 陈艺文 刘钰峰 买尔旦·艾则孜 宋卓霖 潘泓宇 WU Ruijia;Chen Haiyan;CHEN Yiwen;LIU Yufeng;Maierdan·Aizezi;SONG Zhuolin;PAN Hongyu(The Fifth Clinical College of Xinjiang Medical University,Urumqi Xin jiang 8300107;The Fifth Affiliated Hospital of Xinjiang Medical University,Urumqi Xinjiang 8300107)

机构地区：[1]新疆医科大学第五临床医学院,新疆乌鲁木齐8300107 [2]新疆医科大学第五附属医院,新疆乌鲁木齐8300107

出　　处：《当代医药论丛》2023年第7期50-53,共4页

基　　金：新疆维吾尔自治区大学生创新训练计划项目,项目名称:下生殖道微生物群与复发性流产的相关性研究,项目编号:S20220760158。

摘　　要：目的:基于神经网络预测早期先兆流产孕妇的妊娠结局,以期找到一种针对早期先兆流产孕妇妊娠结局的高精度预测方法。方法:以120例先兆流产孕妇的孕酮(P)、雌二醇(E_(2))、人绒毛膜促性腺激素(β-HCG)、孕周和最终妊娠结局为研究样本数据,将P、E_(2)、β-HCG、孕周作为输入向量,将保胎结局作为输出向量,利用学习矢量量化(LVQ)神经网络建立早期先兆流产孕妇妊娠结局预测模型。从研究样本数据中随机提取480份数据作为训练样本,对预测模型进行训练,将其余120份数据作为检测样本输入已训练好的预测模型,检测预测模型的准确率。结果:以P、E_(2)、β-HCG、孕周的定量数据作为输入向量,利用LVQ神经网络对早期先兆流产孕妇的妊娠结局进行预测,结果显示保胎成功预测的正确率为78.43%,保胎失败预测的正确率为94.20%,总体妊娠结局预测的正确率为87.5%,该预测模型具有较高的预测精度。结论:将与妊娠结局具有相关性的P、E_(2)、β-HCG等指标作为输入向量,利用LVQ神经网络建立预测模型,可实现对早期先兆流产孕妇妊娠结局的高精度预测。Objective:To predict the pregnancy outcome of pregnant women with early threatened abortion based on neural network,in order to find a high precision prediction method for pregnancy outcome of pregnant women with early threatened abortion.Method:In 120 pregnant women with threatened abortion,progesterone(P),estradiol(E_(2)),human chorional gonadotropin(β-HCG),gestational age and final pregnancy outcome were taken as sample data.P,E_(2),β-HCG and gestational age were taken as input vectors,and fetal survival outcome as output vectors.The pregnancy outcome prediction model of pregnant women with early threatened abortion was established by using LVQ neural network.480 pieces of data were randomly extracted from the research sample data as training samples to train the prediction model,and the remaining 120 pieces of data were input into the trained prediction model as detection samples to test the accuracy of the prediction model.Results:The quantitative data of P,E_(2),β-HCG and gestational age were used as input vectors to predict the pregnancy outcome of pregnant women with early threatened abortion by LVQ neural network.The results showed that the correct rate of successful prediction of pregnancy protection was 78.43%,the correct rate of failure prediction was 94.20%,and the correct rate of overall pregnancy outcome prediction was 87.5%.The prediction model has high prediction accuracy.Conclusion:Using P,E_(2),β-HCG and other indexes which are correlated with pregnancy outcome as input vector,the prediction model can be established by LVQ neural network,which can achieve the high precision prediction of pregnancy outcome of pregnant women with early threatened abortion.

关 键 词：早期先兆流产 妊娠结局 LVQ神经网络 预测 

分 类 号：R714.21[医药卫生—妇产科学]