输入性甲型H1N1流感疑似病例的调查及分类树判别

The study and predictive model of the classification tree to analyse imported suspected A（H1N1） influenza cases

目的研究输入性甲型H1N1流感流行特征及分类树判别。方法对260例输入性甲型H1N1流感疑似病例进行流行病学描述性研究，用分类树法（CART）建立判别预测模型。结果260例输入性疑似甲型H1N1流感病例中确诊12例，确诊率为4．62％，以中国籍海外留学生为主。单因素分析表明确诊病例与非确诊病例的年龄、性别、职业、腋下体温、红外线监测体温及主要体征和症状（咳嗽、腹泻、关节或肌肉疼痛、鼻塞、自觉咽痛、咽部或扁桃体红肿、流涕）的构成差异无统计学意义（P〉0．05）。CRAT表明判别预测甲型H1N1流感的重要变量依次为年龄、体温、职业、有流涕、咽部或扁桃体红肿和咳嗽，准确率为95．77％。结论甲型H1N1流感疫情与既往流感大流行特征基本类似，CART判别预测有良好的初筛应用价值。

Objective To study the epidemiological characteristics and establish predictive model of the classification tree for imported suspect A（H1N1） influenza cases. Methods Based on the descriptive epidemiological analysis of 260 imported suspected cases, classification and regression tree（CART） was used to predict the A （H1N1） influenza cases. Results In the 260 cases, 12 cases were diagnosed, and the infection rate of A（H1N1） influenza was 4.62%, most of them were Chinese students abroad. The univariate analysis indicated no significant difference between conformed A （H1N1） influenza cases and suspected cases in the distribution of age, gender, occupation, body temperature of mercurial thermometer, surveillance body temperature of infrared rays, cough, diarrhea, joints or muscles pain, nasal obstruction, pharyngodynia, red swelling of the pharynges or tonsils and nasal discharge（ P 〉 0.05）. The significant variables of prediction indicated by CART included age, body temperature, occupation, nasal discharge, pharyngodynia, red swelhng of the pharynges or tonsils and cough. The predictive accuracy ratio was 95.77%. Conclusions The epidemiologieal characteristics of the current A（H1N1 ） influenza pandemic is similar to the previous. The predictive CART model established by outside surveillance factors has the satisfied accuracy and application value for preliminary screening in the quarantine field.