非参数逐步判别分析在脑中风分类诊断中的应用

本文采用不等带宽核密度估计的非参数逐步判别分析,使用SAS软件,对复旦大学附属中山医院78例脑中风样本病例,科学地进行了脑出血、脑缺血的分类诊断。经交叉证实法得判别正确率为:脑出血87 88%、脑缺血88 89%,总判别正确率88 46%,判别效果良好。应用78例训练样本建立的判别函数,在苏州大学附属医院2002年42例脑中风病人的分类诊断中,取得了成功的实际应用效果。

The Application of Classified Diagnosis on Apoplexy Using Stepwise Discriminant Analysis with Nonparametric Method

This paper takes 78 cases of apoplexy in Zhongshan Hospital attached to Fudan University as training samples and diagnoses the case as cerebral haemorrhage or lacking blood in the cerebrum scientifically using stepwise discriminant analysis with nonparametric method. The percentage classified correctly into cerebral haemorrhage is 87.88%,The percentage classified correctly into lacking blood in the cerebrum is 88.89%,the total correct percentage on discrimination is 88.46% and discriminant analysis achieves good results. The discriminant function established by training samples is applied successfully for discriminant diagnosis of 42 cases of aplplexy in the hospital atteched to Suzhou University.