基于非参数交叉证实判别分析法建立社区流行性脑脊髓膜炎初筛诊断模型的研究  

Establishment of Primarily Screening Diagnosis Model for Epidemic Cerebrospinal Meningitis Based on Nonparametric Cross Test Discriminant Analysis

作　　者：刘娟[1] 崔树起[2] 吕曼[2] 王安心[1] 罗艳侠[1] 吴昊[3] 张津[4] 郭秀花[1] 

机构地区：[1]首都医科大学公共卫生与家庭医学学院流行病与卫生统计学系,北京市100069 [2]首都医科大学公共卫生与家庭医学学院全科医学系,北京市100069 [3]首都医科大学北京市佑安医院 [4]首都医科大学北京市宣武医院

出　　处：《中国全科医学》2010年第28期3157-3159,共3页Chinese General Practice

基　　金：教育部人文社会学科研究项目<提高社区医生转诊能力的辅助诊断系统研究>(08JA840023);国家科技支撑计划<社区疾病预防与控制研究>重点项目<社区基层卫生服务信息技术研究>(2007BAI24B07)

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。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.

关 键 词：非参数判别分析 交叉证实 诊断模型 流行性脑脊髓膜炎 

分 类 号：R515.2[医药卫生—内科学] R197[医药卫生—临床医学]