ARIMA模型在常见呼吸道传染病疫情预测中的应用

Application of ARIMA model in prediction of respiratory infectious disease

目的探讨应用时间序列求和自回归移动平均(ARIMA)模型预测本地区常见呼吸道传染病的发病情况。方法利用《中国疾病预防控制信息系统》的资料,应用SPSS 17.0统计软件、采用ARIMA模型,对合肥市蜀山区呼吸道传染病逐月的发病情况进行建模和拟合,利用所得到的模型对2012年各月发病情况进行预测,并评价其预测效果。结果合肥市蜀山区6种常见呼吸道传染病的季节性表现为每年3～5月份和每年的11月～次年的1月份为高发月。ARIMA(0,0,1)(0,0,1)是本地区常见呼吸道传染病拟合的最佳模型,模型拟合统计量均方根误差(RMSE)为20.299,平均绝对百分位误差(MAPE)为41.264,正态化的贝叶斯信息准则(BIC)为6.226,Ljung-Box Q值为0.375,即P>0.05,可知残差属于白噪声值。结论 ARIMA模型对蜀山区常见呼吸道传染病拟合的预测效果较为满意,预测结果将为今后常见呼吸道传染病的预防和控制提供理论支持。

Objective To explore the application of time series autoregressive integrated moving average （ ARIMA ）model for the prediction of respiratory infectious disease in Shushan area. Methods Based on the data gathered in Natio, nal Disease Reporting Management System, the prevalence of respiratory infectious diseases in Shushan district of Hefei city was studied by modeling and data fitting with the help of statistical software SPSS 17.0 and ARIMA model. The established model was used to forecast monthly prevalence of the respiratory infectious diseases in 2012, and the results were evaluated at the same time. Results Two periods （from March to May and from No- vember to January of the following year） had the higher occurrence rate for 6 kinds of common respiratory infectious diseases in Shushan district of Hefci city. ARIMA （0,0,1）was determined to be the optimal model for fitting of common infectious diseases, and root mean square error （RMSE） was 20. 299, mean absolute percentage error （MAPE） was 41. 264, the normalized bayesian information criterions （BIC） was 6. 226, and the value of Ljung- Box Q was 0. 375 （P 〉 0. 05 ）. The above fitting results showed that the residual error of the model accorded with the white noise standard. Conclusion The forecasting results of the common infectious diseases in Shushan district based on ARIMA model fitting is satisfactory, which can provide theoretical support for common infectious diseases prevention and control of our area in future.