Logistic回归判别模型判别HIV抗体不确定者转归的可行性  

The feasibility of logistic regression model in determining outcome of HIV western blot-indeterminate patients

作　　者：陈慧萍[1] 刘雷[2] 张连生[2] 彭国平[1] 蒋洪林[1] 汤恒[1] 彭庭海[1] 阳凯[1] 

机构地区：[1]湖北省疾病预防控制中心,武汉430070 [2]武汉科技大学医学院公共卫生学院,武汉430065

出　　处：《中国艾滋病性病》2014年第5期367-369,379,共4页Chinese Journal of Aids & STD

基　　金：2012年湖北省公益性科技研究项目(2012DCA14001)~~

摘　　要：目的探讨用Logistic回归判别模型判别艾滋病病毒(HIV)抗体不确定者转归的可行性。方法对2002-2013年被诊断为HIV抗体不确定者,按照《全国艾滋病检测技术规范》进行随访检测。描述分析HIV抗体不确定者的流行病学特征,用Logistic回归模型判别分析HIV抗体不确定者的转归。结果共有162例受检者被判定为"HIV抗体不确定"并成功随访,其中44例随访检测阳性,阳转率为27.16%(44/162)。Logistic回归判别模型的特异度、灵敏度分别为96.61%(114/118)、70.45%(31/44),总判别准确率为89.51%(145/162)。结论模型判别准确率良好,提示Logistic回归判别模型可以作为HIV的辅助诊断方法之一。Objective To investigate the feasibility of logistic regression discriminant model in determing outcome of human immunodeficiency virus （HIV） western blot-indeterminate （WBi） patients. Methods Accoring to national guideline for detection of HIV/AIDS, subjects with WBi were followed up from 2002 to 2013. Epidemiological characteristics of HIV indeterminate results were analyzed. Logistic regression was built to determine outcome of HIV indeterminate results. Results A total of 162 subjects were diagnosed as HIV western blot-indeterminate and successfully followed up. Among them, 44 turned positive and the positive rate was 27. 16% （44/162）. The specificity, sensitivity and total accuracy of the logistic regression model were 96.61 % （114/118）, 70.45 % （31/44） and 89.50 % （145/162）. Conclusion The logistic regression model has good prediction accuracy, suggesting that the model can be used in diagnosis of HIV infection.

关 键 词：HIV抗体不确定者 Logistic回归判别模型 判别分析 

分 类 号：R373.9[医药卫生—病原生物学]