用时间序列分析对黄浦区流感样病例就诊百分比建模并评价预测效果

The application of time series analysis in modeling and predicting the influenza-like illness consultation rate in Huangpu District of Shanghai

目的利用时间序列分析对黄浦区流感样病例(ILI)就诊百分比建模并评价预测效果。方法通过国家流感监测信息系统收集黄浦区2011年第1周—2014年第52周的ILI就诊百分比,利用SAS 9.3软件对数据建立最优ARIMA模型,利用2015年第1周—第6周的ILI就诊百分比对已建立的ARIMA模型进行预测效果评价。结果对ILI就诊百分比预处理发现序列为平稳非白噪声序列,最终建模为ARIMA(1,0,0),模型残差为白噪声序列,真实值均在预测值95%可信区间内。结论 ARIMA(1,0,0)模型可用于黄浦区ILI就诊百分比的预测。

Objective To explore the feasibility for modeling and forecasting outpatient consultationrate of influenza-like illness（ ILI） in Huangpu District of Shanghai. Methods ILI consultation rates from the first week of 2011 to the 52 th week of 2014 were collected through the National Influenza Surveillance Information System. SAS 9. 3 was used to establish an optimal ARIMA model of ILI data. The forecasting ability of the ARIMA model was afterwards evaluated by using data from the first to the sixth week in 2015. Results ARIMA（ 1,0,0） was identified as the final model in fitting outpatient vising rate of ILI with stable and white noise residual. Model validation showed that the real outpatient-visiting rate of ILI all fell in the 95% confidence interval of the predictive values. Conclusion ARIMA（ 1,0,0） could be used in the forecast of ILI consultation rates in Huangpu District of Shanghai.