支持向量机在胎膜早破预测中的应用

为快速、准确地对胎膜早破进行预测,首次应用了一种新型的数据挖掘技术-支持向量机预测模型.该模型针对所获取的胎膜早破及正常破膜数据集100个病例进行建模,并与神经网络、Logistic回归建模的性能进行了比较.结果表明,支持向量机具有可调参数少、学习速度快等优点,计算所得到的结果无论从准确率,还是所获取知识的可理解性等方面,都优于常用的神经网络等方法.用支持向量机方法建立的胎膜早破预测模型合理可行.

Application of Square Support Vector Machine to Predict Premature Rupture of Fetal Membranes

A new type data mining technique,Support Vector Machine(SVM),is firstly applied to perform fast and accurate prediction classification of premature rupture of fetal membranes(PROMs) based on 100 PROM and nonPROM cases.The prediction ability of SVM model is compared with BP network and logisitc regression.The results show that SVM has only two parameters and can be trained fastly.More important,SVM is capable of getting higher prediction accuracy and reliability than BP network and logisitc regression.In conclusion,PROM prediction model established with SVM is reasonable and practicable.