江西省肾综合征出血热发病率ARIMA模型及其趋势预测  

Application of autoregressive integrated moving average model in incidencerate prediction of hemorrhagic fever with renal syndrome in Jiangxi province

作　　者：朱蒙曼 杨其松 谢昀[2] 张天琛[2] 刘美 刘晓青[2] 

机构地区：[1]南昌大学公共卫生学院流行病与卫生统计学教研室,江西南昌330000 [2]江西省疾病预防控制中心应急办与传染病防制所,江西南昌330029

出　　处：《现代预防医学》2017年第22期4036-4041,共6页Modern Preventive Medicine

基　　金：江西省青年科学基金资助项目(20171BAB215051)

摘　　要：目的探讨自回归求和移动平均模型(Auto Regressive Integrated Moving Average,ARIMA)在江西省肾综合征出血热月发病率预测的可行性,为制定出血热防控策略提供依据。方法基于江西省2006-2015年肾综合征出血热(Hemorrhagic Fever with Renal Syndrome,HFRS)逐月发病率资料建立ARIMA模型,利用2016年各月发病率检验模型预测效果,再以2006-2016年HFRS逐月发病率构建模型预测2017年HFRS发病率。结果本研究构建的ARIMA(0,0,2)(0,1,1)12模型,拟合结果与实际发病情况基本吻合。各项参数均有统计学意义(P<0.05),BIC值(Schwarz Bayesian criterion,贝叶斯信息准则)=-6.792,Ljung-Box Q=14.992,P=0.452,模型残差为白噪声;2016年各月HFRS发病率预测值与实际值动态趋势基本吻合。预测2017年江西省HFRS发病率为1.45/10万。结论 ARIMA模型能很好地模拟江西省HFRS发病率在时间序列上的变动趋势,可用于江西省HFRS发病率的短期预测研究。Objective The aim of this study was to establish an autoregressive integrated moving average （ARIMA） model and explore the application of this model for epidemic trend prediction of hemorrhagic fever with renal syndrome （HFRS） in Jiangxi province. Methods ARIMA model was established based on the monthly incidence rate of HFRS in Jiangxi province from 2006 to 2015. The constructed optimal model was applied to predict the incidence rate of HFRS in 2016, and to evaluate the validity of the model through comparing the difference of predicted incidence rate and actual one. The incidence rate of HFRS in 2017 was predicted by the ARIMA model based on the incidence rate from 2006 to 2016. Results The model of ARIMA （0,0,2） （0,1,1） 12 had a good fitness to the incidence rate, and all parameters of the model were statistically significant （P〈0.05） with Schwarz Bayesian criterion （BIC）=-6.792. The residual error was white noise, and the Ljung-Box test statistics for the model was 14.992 with a P value of 0.452. The predicted incidence rate in 2016 was consistent with the actual one. The predicted incidence rate of HFRS in 2017 was 1.45 per billion. Conclusion ARIMA model can be used to simulate changes in HFRS incidence rate and to forecast the incidence rate in short-term in Jiangxi province.

关 键 词：时间序列 ARIMA模型 肾综合征出血热 预测 

分 类 号：R181.2[医药卫生—流行病学]