胎儿体重_预测_神经网络(生物医学信息)

根据孕妇参数预测胎儿体重的神经网络方法

http://www.100md.com 《中国生物医学工程学报》 1999年第2期

胎儿体重|预测|神经网络,关键词：

     刁晓娣¹ 江志斌² 刘瑾¹ 刁晓娣刘瑾安徽医科大学附属医院妇产科，合肥 230022；江志斌合肥工业大学生物机电所，合肥 230069 中国生物医学工程学报 1999 0 18 2

    关键词：胎儿体重；预测；神经网络期刊 zgswyxgcxb 0 生物医学信息 fur -->

本文采用反向传播神经网络算法，根据孕妇身高、体重、宫高及腹围预测胎儿体重。建立了一个预测胎儿体重的网络模型，讨论了确定网络拓扑结构的方法。采用该方法预测了140例胎儿体重，预测符合率高达85%，相对误差10%者占预测总数的94.28%。采用神经网络分析输入对于输出的贡献的结果表明孕妇宫高对于胎儿体重影响最大。

分类号： R318.04; R714.51

AN ARTIFICIAL NEURAL NETWORK BASED ESTIMATIONOF FETAL WEIGHT BY USING PREGNANTPARAMETERS

Diao Xiaodi¹, Jiang Zhibin²,Liu Jin¹

(1 Anhui Medical University, Hefei 230022； 2 Hefei University ofTechnology, Hefei 230069)

ABSTRACT ：Artificial NeuralNetwork (ANN) was applied to estimate fetal weight based on height, weight, height ofuterine fundus and abdominal perimeter of pregnant women. An ANN model for estimatingfetal weight was established, and method for determining topological structure of themodel was discussed. Furthermore, by this ANN model, weights of 140 fetus were estimated.The result of the estimation was that the correspondence rate was as high as 85%, and thecases with relative error10% made up 94.28 of the estimated cases. The ANN analysis onthe contribution of the input to the output showed that the height of uterine fundus hadthe largest effect on the fetal weight.

......

百拇医药网 http://www.100md.com/html/DirDu/2005/07/04/66/77/30.htm

您现在查看是摘要页，全文长 17949 字符。