胎盘植入的BP神经网络辅助诊断模型  

Auxiliary Diagnosis Model of Placenta Implantation of BP Artificial Neural Networks

作　　者：杨丹林[1] 何斌杰 张栋[2] 颜建英[1] 

机构地区：[1]福建省妇幼保健院妇产科,福建福州350001 [2]福州大学数学与计算机科学学院,福建福州350116

出　　处：《中国卫生标准管理》2017年第28期37-38,共2页China Health Standard Management

基　　金：福建省科技计划引导性项目(2016Y0060)

摘　　要：目的通过部分特征数据,利用人工神经网络技术,判断患者是否存在胎盘植入高危因素和患有胎盘植入,从而得到一个临床预测辅助诊断模型。方法利用人工神经网络技术,设计一个三层BP神经网络模型,将14个特征值作为神经网络的输入数据,将是否为胎盘植入作为输出数据。结果共收集266条孕产妇检测数据(其中226条为训练样本,40条为测试样本),抽取出14个有可能影响胎盘植入并发症的特征指标如血清白蛋白、宫颈长度、胎盘位置、流产次数等。训练好的神经网络经测试组测试判断正确率为85%。结论可以利用训练好的神经网络作为胎盘植入的临床预测辅助诊断模型。Objective To determine high risk factors of patients with placenta implantation through some characteristic data and artifcial neural network technology, and obtaine a clinical prediction auxiliary diagnosis model. Methods Artificial neural network technology was proposed. It designed a three layers BP neural network model, which included the 14 characteristic value as the input data of neural network for placenta implantation as output data. Results It collected 266 pregnant test data （including 226 training samples, 40 for test samples）, 14 may affect implanted placenta extract the characteristic indices such as serum albumin of complications, cervical length, position of the placenta, abortion times, etc. Trained neural network by the test group test judgment accuracy was 85%. Conclusion The trained neural network was taken advantage as an aid in the diagnosis of placenta increta clinical prediction model.

关 键 词：胎盘植入 BP神经网络 辅助诊断模型 

分 类 号：TP391.77[自动化与计算机技术—计算机应用技术]