应用人工神经网络预测再次妊娠肝内胆汁淤积症孕妇发生羊水浑浊的风险

目的:构建再次妊娠妊娠期肝内胆汁淤积症(ICP)孕妇羊水混浊的预测模型,探讨相关指标的预测价值。方法:收集2009年1月至2014年8月在浙江大学医学院附属妇产科医院再次妊娠住院分娩的ICP孕妇的临床资料及胎儿风险相关资料,应用人工神经网络构建羊水混浊预测模型,分析相关指标对羊水混浊的预测结果和影响权重。结果:应用人工神经网络模型预测ICP孕妇羊水混浊灵敏度为68.0%,特异性为85.0%,准确率为80.3%。参数权重在10%以上的因素有妊娠合并症、孕妇分娩前血清甘胆酸浓度和孕妇年龄。结论:人工神经网络可用于构建ICP孕妇胎儿宫内环境即羊水混浊的预测模型;影响再次妊娠ICP孕妇胎儿宫内安全的危险因素有孕妇年龄、妊娠合并症、孕妇分娩前血清甘胆酸浓度等。
Abstract: Objective: To establish a prediction model of fetal meconium-stained amniotic fluid in re-pregnant women with intrahepatic cholestasis of pregnancy(ICP). Methods: Clinical data of 180 re-pregnant women with ICP delivering in Women's Hospital, Zhejiang University School of Medicine between January 2009 to August 2014 were collected. An artificial neural network model(ANN) for risk evaluation of fetal meconium-stained fluid was established and assessed. Results: The sensitivity, specificity and accuracy of ANN for predicting fetal meconium-stained fluid were 68.0%, 85.0% and 80.3%, respectively. The risk factors with effect weight >10% were pregnancy complications, serum cholyglycine level,maternal age. Conclusion: The established ANN model can be used for predicting fetal meconium-stained amniotic fluid in re-pregnant women with ICP. Key words: Neural networks(computer) Pregnancy Fetal monitoring Cholestasis, intrahepatic Amniotic fluid Forecasting