中文摘要：

目的：建立适用于社区中脑膜炎与其他中枢神经系统疾病的鉴别诊断模型。方法：采用不等带宽核密度估计的非参数判别分析，对中国典型病例大全近四年内符合纳入标准的161例脑膜炎和161例非脑膜炎患者完整的病例资料进行分析。结果：经交叉证实法得到脑膜炎组的判别正确率为83．95％，对照组为71．25％，总的判断正确率87．64％。同时对资料进行logistic回归和人工神经网络模型进行分析，并进行与人工神经网络和logistic回归所建立的模型进行比较。结论：非参数判别分析建立的脑膜炎诊断模型是理想模型。

英文摘要：

Objective： To get a differential diagnosis model for meningitis and other central nervous system diseases depended on 3-5 clinical symptoms and signs. Method： Analysis 161 cases of meningitis and 161 cases of non-meningitis cases of patients with complete information from full-text databases of China National Knowledge Infrastructure （CNKI） with discriminant analysis method. Results： The percentage classified correctly into meningitis is 87. 88%, the percentage classified correctly into non-meningitis is 88. 89%, and the total correct percentage on discrimination is 88. 46 %. The data was analyzed by logistic regression and artificial neural networks too. Conclusion：The research shows that nonparametric discriminant analysis is better.