时间序列分析在甲型肝炎疫情预测中的应用  

Application of time series analysis in forecasting the epidemic trend of hepatitis A

作　　者：陶长余[1] 陈郁[1] 章士军[1] TAO Chang- yu CHEN Yu ZHANG Shi- jun(Nantong City Center for Disease Control and Prevention, Jiangsu 226007, Chin)

机构地区：[1]南通市疾病预防控制中心,江苏226007

出　　处：《医学动物防制》2016年第10期1113-1115,共3页Journal of Medical Pest Control

摘　　要：目的应用季节乘积求和自回归移动平均模型分析南通市甲型肝炎（简称甲肝）每月发病数时间序列,建立预测模型。方法收集南通市2009年1月-2015年9月间甲肝病例月报告数据,应用EVIEWS软件拟合ARIMA模型,最后进行预测分析。结果成功建立模型ARIMA[（2）,0,（2）],模型表达式为：xt=8.4 419＋（1＋0.6 182 B2）t/（1-0.7 474B2）,模型通过参数检验及残差白噪声检验（P〉0.05）。预测2015年4月-2015年9月发病数,平均相对误差为30.17%,模型拟合效果较好。预测2015年10月-2016年3月发病数,显示发病趋势较为平稳。结论求和自回归移动平均模型对南通市甲肝发病情况拟合和趋势预测效果较好,可根据预测结果开展甲肝疫情相关防控工作。Objective The purpose of this study was to establish the predictive model by using ARIMA model to analyze monthly data of hepatitis A cases in Nantong city. Methods Firstly, the data of hepatitis A cases between 2009 Jan. and 2015 Sep in Nantong city was collected, secondly, the Eviews software was used to fit ARIMA model, and finally, the optimal model was used to forecast and analyze. Results The model ARIMA [ （2）, 0, （2） ] was established, and the formula was xt = 8.4 419 ＋ （1 ＋0. 6 182 B2 ）εt/（ 1 -0. 7 474B2 ）, the parameters had passed the significant test, and the residual series of the model had passed the white noise test（P 〉 0. 05）. It showed that the model fitted better with the average relative error of 30. 17% between the actual number and the predicted number during 2015 - 04 -2015 - 09. And the epidemic trend was relatively stable during 2015 - 10 - 2016 -03 predicted by the model. Conclusion The model better fitted the time series of hepatitis A cases in Nan- tong City2. Relevant prevention measures should be taken against the hepatitis A prevailing according to the forecast results.

关 键 词：求和自回归移动平均模型 时间序列 甲型肝炎 预测 

分 类 号：R512.61[医药卫生—内科学]