支持向量机对艾滋病病毒感染者初筛识别的研究  

A HIV Carrier Forecasting Model for Quarantine Based on Support Vector Machines

作　　者：吴海磊[1] 钱吉生[1] 张纯[1] 丁永健[2] 陆永昌[2] 尹军[2] 

机构地区：[1]南京出入境检验检疫局卫生与食品检验监督处,江苏南京211106 [2]江苏出入境检验检疫局卫生监管处

出　　处：《实用预防医学》2010年第11期2152-2155,共4页Practical Preventive Medicine

基　　金：国家质量监督检验检疫总局科研项目(2009IK215)

摘　　要：目的在描述性分析的基础上,应用支持向量机（support vector machines,SVM）建立艾滋病病毒（human im-munodeficiency virus,HIV）感染者的识别模型。方法以2004-2009年某省口岸在出入境人员中检出的133例HIV感染者及非HIV感染者133例组成研究对象,进行流行病学描述性分析。以是否感染HIV为因变量,以年龄、性别、国籍、文化程度、职业、劳务史、会阴部症状史、不洁性生活史、吸毒史、输血史、性伴侣是高危人员、是否有同性性行为、性伴侣数为自变量建立SVM模型,分析预测效果和预测变量的重要性。结果 HIV感染者以35-44岁年龄为主（48.87%）,男性占93.23%,劳务人员占48.12%,性伴侣2人及以上的占51.88%,承认不洁性生活史占39.10%,承认有同性性行为占3.76%。SVM模型识别总体准确率为90.60%（灵敏度90.29%,特异度90.90%）。预测变量重要程度以劳务史为首（0.80）,其次是不洁性生活史（0.04）和性伴侣是高危人群（0.03）。结论口岸HIV感染者有其特定的流行病学特征,HIV感染者SVM识别模型的拟合能力强,能较好地用于评价口岸现场未知样本的初筛。Objective To construct a HIV carrier forecasting model based on support vector machines（SVM）. Methods The epidemiological features of all 133 HIV carriers found in the port of a province in China during the period of 2004~2009 were analyzed.The data of another 133 individuals without HIV were also used for building the forecasting model.First,the epidemiological features of these HIV carriers were analyzed.Then the forecasting model based on SVM for HIV carriers were constructed.The independent variables included age,gender,nationality,educational level,occupation,history of labor,perineal position symptoms,sexual activity without protection,sexual partner with homosexuality,sexual partner with high risk factors of sexual transmitted diseases,the number of sexual partner, and history of drug taking or blood transfusion.Third,the effect of forecasting model and the importance of all independent variables were analyzed. Results Most of 133 HIV carriers were male（93.23%）,whose ages were from 35 to 44 years（48.87%）.The main occupation was laborer（48.12%）.51.88% of them acknowledged that the number of their sexual partner was not less than 2.39.10% acknowledged that they had sexual activity without protection,and 3.76% acknowledged that they had homosexual sex.The overall accurate rate of forecasting of this model was 90.60%.Its sensibility and specificity were 90.29% and 90.90%,respectively.History of labor,unclean sexual intercourse and sex partner with high risk factors of sexual transmitted diseases were the most important influencing factors. Conclusions The epidemiological features of HIV carriers found in the frontier port？have special characteristics and the model of HIV carriers forecasting based on SVM has a good fitting ability,which could be used for predicting the unknown sample effectively.

关 键 词：HIV感染者 支持向量机 预测 流行病学 

分 类 号：R512.91[医药卫生—内科学]