中文摘要：

目的建立适用于社区的流行性脑脊髓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,对北京市佑安医院近五年内符合纳入标准的106例流行性脑脊髓膜炎患者和北京市宣武医院近五年内符合纳入标准的106例其他中枢神经系统疾病患者完整的病例资料进行分析。结果经交叉证实法得到发热、头痛、呕吐、颈强直和瘀点瘀斑建立的流行性脑脊髓膜炎诊断五联征,灵敏度为：100.00%,特异度为：83.53%,阳性预测值为：85.86%,阴性预测值为：100.00%,调整一致性为：92.35%。结论发热、头痛、呕吐、颈强直和瘀点瘀斑建立的流行性脑脊髓膜炎诊断模型有较好的识别作用。

英文摘要：

Objective To set up a differential diagnosis model for epidemic cerebrospinal meningitis（ECM） and other central nervous system diseases,which is suitable for application in community.Methods Nonparametric discriminant analysis of unequal bandwidth kernel density estimation was used to analyze the data of 106 ECM patients who received medical care in Bejing You′an Hospital and another 106 patients with other central nervous system diseases who received medical care in Beijing Xuanwu hospital in recent 5 years,and all the cases met the inclusion criteria.Results The diagnostic model for ECM on the basis of fever,headache,vomiting,neck stiffness and rash was obtained though cross test.Its sensitivity was 100.00%,specificity was 83.53%,positive predictive value was 85.86%,negative predictive value was 100.00%,and adjusted consistency was 92.35%.Conclusion The diagnosis model of ＂fever,headache,vomiting,neck stiffness and rash＂ had a good diagnostic value.