应用时间序列法建立学校结核病预警模型研究  

Time series analysis for early warning model of tuberculosis in schools

作　　者：刘维华[1] 朱建良[1] 郭付爱[1] 张洁[1] LIU Weihua;ZHU Jianliang;GUO Fuai;ZHANG Jie(Shijiazhuang Center for Disease Control and Prevention,Shifiazhuang (050011),China)

机构地区：[1]河北省石家庄市疾病预防控制中心结核病防治所,050011

出　　处：《中国学校卫生》2018年第11期1682-1685,共4页Chinese Journal of School Health

摘　　要：目的建立石家庄地区学校结核病预警模型,为学校早期甄别和科学控制结核病疫情提供参考。方法从"结核病信息管理系统"收集2012—2017年石家庄地区学校每月总体学生报告病例数和非健康检查学生报告病例数,进行季节性分解,按照指数平滑法或自回归移动平均模型(ARIMA)方法进行最优化模型筛选。利用2012—2017年石家庄地区学校结核病聚集性疫情的数据验证模型,并确定对疫情进行预警的阈值条件。结果对于总体学生发病例数数据拟合度最好的是ARIMA(0,0,0)(1,1,0),平稳R^2、R^2和正态化BIC值分别为0.71,0.61,4.70;对于非健康检查学生发病例数数据,拟合度最好的是ARIMA(0,1,1)(1,0,0),平稳R^2、R^2和正态化BIC值分别为0.60,0.54,4.08。如果当月非健康检查学生病例预测模型观察值高于拟合值,并且总体学生病例预测模型中,观察值也高于拟合值,提示学校结核病聚集性疫情风险增大。对非健康检查学生病例预测模型采用逐一预测模式,对总体学生病例预测模型采用连续预测模式。结论利用时间序列法建立的预警模型有助于发现学校结核病疫情。Objective To establish early warning model of tuberculosis (TB)based in schools of Shijiazhuang.Methods The monthly reported TB cases in schools in Shijiazhuang from 2012to 2017were collected from Tuberculosis Information Management System.Student TB cases per month from routine as well as random monitoring approaches were collected.The optimal model was selected according to the exponential smoothing method or ARIMA method.By comparing data on reported TB cases from schools during 2012-2017,the threshold of early warning for school tuberculosis cluster cases were verified.Results The best model for school TB cases was the simple seasonal model with the smooth R^2,R^2and normalized BIC at 0.711,0.608,and 4.70, respectively.For school TB cases from non-health screening approaches,the best model was Winters plus model with smooth R^2, R^2,and normalized BIC at 0.597,0.544,and 4.082,respectively.When the observed value of Winters plus model was higher than the fitted value,and the observed value of simple seasonal model was also higher than the fitted value,it was the threshold of early warning of school TB outbreak which indicated the enhanced risk.One by one prediction and series prediction were used in simple seasonal model and Winters plus model,respectively.Conclusion The early warning models established by time series analysis is helpful for the early detection of TB epidemic in schools.

关 键 词：时间 序列分析 结核 学生保健服务 

分 类 号：R181.8[医药卫生—流行病学] R52[医药卫生—公共卫生与预防医学]