应用ARIMA模型对河南省肺结核流行趋势预测分析  

Applying ARIMA model to predict the epidemic trend of pulmonary tuberculosis in Henan

作　　者：蒋建国[1] 孙定勇[1] 张艳秋[1] 何梦雅 庄严[1] 徐吉英[1] 孙建伟[1] JIANG Jianguo;SUN Dingyong;ZHANG Yanqiu;HE Mengya;ZHUANG Yan;XU Jiying;SUN Jianwei(Institute for Tuberculosis Control and Prevention,Henan Center for Disease Control and Prevention,Zhengzhou,Henan 450016,China)

机构地区：[1]河南省疾病预防控制中心结核病预防控制所,河南郑州450016

出　　处：《现代疾病预防控制》2023年第7期495-499,563,共6页MODERN DISEASE CONTROL AND PREVENTION

基　　金：河南省科技发展计划(212102310156);河南省医学科技攻关计划联合共建项目(LHGJ20200131)。

摘　　要：目的采用自回归移动平均模型法(autoregressive integrated movingaverage,ARIMA)构建时间序列模型,对2007-2022年河南省肺结核流行趋势进行分析,并进行短期预测。方法从中国疾病预防控制中心《中国结核病信息管理系统》收集2007-2022年河南省肺结核月登记率数据,应用SPSS 27.0软件的时间序列预测模块,构建时间序列模型,验证模型效果,对2023-2025年肺结核登记率进行预测。结果以2007-2019年河南省肺结核月登记率构建最优模型,ARIMA(0,1,2)(1,0,0)12,Ljung-Box统计量Q=13.922,P=0.531,平稳R2为0.734。以2020-2022年实际月登记率作为验证数据集,实际值与预测值拟合良好(平稳R2=0.734)。应用最优模型预测2025年河南省肺结核年登记率将达到18.5/10万,比2015年下降72.1%。结论以2007-2019年肺结核月登记率ARIMA(0,1,2)(1,0,0)12模型能较准确地预测河南省肺结核近期流行趋势,但新型冠状病毒感染(新冠感染)疫情对肺结核的登记率可能存在一定影响。2023-2025年肺结核登记率预测需考虑新冠感染后时代影响因素。Objective To construct a time series model to analyze the epidemic trend of pulmonary tuberculosis in Henan from 2007 to 2022,and make a short-term prediction.Methods The monthly registration rate data of tuberculosis in Henan province from 2007 to 2022 were collected from China Tuberculosis Information Management System of China CDC,and the time series prediction module of SPSS 27.0 was used to build a time series model to verify the effect of the model and predict the registration rate of pulmonary tuberculosis from 2023-2025.Results The optimal model was constructed based on the monthly registration rate of pulmonary tuberculosis in Henan province from 2007 to 2019.ARIMA(0,1,2)(1,0,0)12,Ljung-Box statistic Q=13.922,P=0.531,and stable R?was 0.734.Taking the actual monthly registration rate from 2020 to 2022 as the validation data set,the actual value and the predicted value fit well(stable R2=0.734).The optimal model predicts that the annual registration rate of pulmonary tuberculosis in Henan province will reach 18.5/10'in 2025.Conclusions The ARIMA(0,1,2)(1,0,0)12 model with the monthly registration rate of pulmonary tuberculosis from 2007 to 2019 can accurately predict the recent epidemic trend of pulmonary tuberculosis in Henan,but the epidemic situation of COVID-19 infection has a certain impact on the registration rate of pulmonary tuberculosis.

关 键 词：自回归移动平均模型 肺结核 预测 新型冠状病毒感染 

分 类 号：R52[医药卫生—内科学]