应用ARIMA乘积季节模型预测青岛市甲肝发病  

Application of multiple seasonal ARIMA model in forecasting incidence of hepatitis A in Qingdao

作　　者：梁纪伟[1] 姜法春 韩雅琳[1] 薛爱丽 宋鑫[1] 

机构地区：[1]青岛市疾病预防控制中心,山东青岛266000 [2]青岛市妇女儿童医院

出　　处：《中国公共卫生管理》2016年第6期780-782,793,共4页Chinese Journal of Public Health Management

基　　金：青岛市科技指导计划项目(KJZD-14-19-NSH)

摘　　要：目的探索应用自回归求和移动平均(autoregressive integrated moving average,ARIMA)乘积季节模型分析时间序列对青岛市甲型病毒性肝炎(简称甲肝)发病进行预测分析,为地区甲肝防治提供决策依据。方法利用青岛市自1996以来19年甲肝疫情数据,建立青岛市甲肝ARIMA乘积季节预测模型,并以2015年第一季度月发病数评估模型预测效果。结果季节自回归参数为0.2639,t检验P<0.01,AIC=370.2511,SBC=392.6405,模型残差白噪声χ~2检验P值均>0.05,建立ARIMA乘积季节模型ARIMA(1,1,1)(1,1,2)12,模型表达式(1-0.2639L)(1+1.1814L^(12))(1-L)(1-L^(12))LY=(1+0.1625L^(12)-1.2344L^(24))ε_t,以此开展甲肝发病数预测。结论 ARIMA乘积季节模型能够较好地模拟青岛市甲肝发病趋势,可用于短期预测该地区甲肝发病,为疫情防控提供一定的科学依据。Objective To forecast the incidence of hepatitis A in Qingdao by multiple seasonal autoregressive integrated moving average（ARIMA） model of time series,and to provide decision basis for local prevention and control of hepatitis A. Methods A multiple seasonal ARIMA model was fitted with data of monthly reported cases in Qingdao from 1996 to2014 for hepatitis A and predicted data in the first season of 2015. Results Seasonal auto-regressive coefficient was0.2639. P valves for t-test of all coefficients were all below 0.01.AIC=370.2511, SBC=392.6405.Autocorrelation test for residuals of model was white-noise series（P〈0.05）. ARIMA（1,1,1）（1,1,2）12was identified to fit and forecast monthly hepatitis A cases. Model equation was（1-0.2639L）（1 ＋1.1814L^12）（1-L）（1-L^12） LY =（1 ＋0.1625L^12-1.2344^624）ε_t.Conclusion The multiple seasonal ARIMA model can be used to fit trends for incidence of hepatitis A in Qingdao and forecast the incidence within a short period, and it can provide certain scientific basis for prevention and control of hepatitis A.

关 键 词：甲肝 ARIMA乘积季节模型 时间序列 预测 

分 类 号：R183.4[医药卫生—流行病学] R181.8[医药卫生—公共卫生与预防医学]