应用Python编程语言构建宁波市水痘发病率预测的季节性ARIMA模型  

Establishing a seasonal ARIMA prediction model of varicella incidence in Ningbo city using Python programming language

作　　者：潘兴强[1] 马瑞[1] 杨天池[1] 陈奕[1] 丁克琴[1] 许国章[1] Pan Xingqiang;Ma Rui;Yang Tianchi;Chen Yi;Ding Keqin;Xu Guozhang(Ningbo Municipal Center for Disease Control and Prevention,Ningbo 315010,Zhejiang,China)

机构地区：[1]宁波市疾病预防控制中心,浙江宁波315010

出　　处：《中国疫苗和免疫》2022年第1期83-87,104,共6页Chinese Journal of Vaccines and Immunization

基　　金：浙江省基础公益研究计划项目(LGF20H260006);宁波市科学技术局科技计划项目(202002N3185);宁波市医疗卫生品牌学科(PPXK2018-10)。

摘　　要：目的构建水痘发病率预测的季节性求和自回归移动平均(Autoregressive integrated moving average,ARIMA)模型。方法通过中国疾病预防控制信息系统收集宁波市2008年1月-2021年5月各月水痘发病率数据,按照AIC和BIC准则拟合最优的季节性ARIMA(p,d,q)×(P,D,Q)s模型,对模型进行参数估计,对2021年6-12月各月水痘发病率进行预测。结果2008年1月-2021年5月宁波市水痘报告发病率总体呈上升趋势,其中2008年1月-2013年12月呈快速上升趋势,2014年1月-2019年12月呈缓慢上升趋势。拟合的最优模型为季节性ARIMA(1,1,2)×(1,1,1)12,AIC和BIC值分别为464.7和575.7,模型残差经Ljung-Box检验为白噪音(P=0.94),2010年1月-2021年5月水痘报告发病率的模型预测值与实际发病率流行趋势一致,模型拟合效果较好。模型预测2021年6-12月的水痘发病高峰为10-12月。结论拟合的季节性ARIMA(1,1,2)×(1,1,1)12模型能较好地预测宁波市水痘发病率趋势。Objective To establish an autoregressive integrated moving average(ARIMA)model to predict the incidence of varicella.Methods We obtained varicella incidences between January 2008 and May 2021 in Ningbo city from the China Disease Prevention and Control Information System.We fit a seasonal model of ARIMA(p,d,q)×(P,D,Q)s according to the AIC and BIC criteria,conducted parameter estimation of the fitted model,and predicted the incidence of varicella from June to December in 2021.Results The reported incidence of varicella increased from 2008 to 2021 in Ningbo,with a rapid increase between 2008 and 2013 and a slower increase from 2014 to 2019.The fitted optimal seasonal model was ARIMA(1,1,2)×(1,1,1)12 with AIC and BIC values being 464.7 and 575.7,respectively.The Ljung-Box test for model residuals showed white noise(P=0.94).Predicted varicella incidences from January 2010 to May 2021 using the fitted model were consistent with actual incidences,indicating a good fit of the model.The predicted peak incidence of varicella for June to December 2021,using the model,was between October and December.Conclusions The fitted seasonal model of ARIMA(1,1,2)×(1,1,1)12 better predicted the varicella incidence pattern in Ningbo.

关 键 词：水痘 发病率 预测 求和自回归移动平均模型 Python编程语言 

分 类 号：R186[医药卫生—流行病学]