时间序列分析在上海静安区流感样病例预测预警中的应用
Application of Time Series Analysis in Forecasting and Early Warning of Influenza-Like Illness in Jing'an District, Shanghai
-
摘要:
目的 探讨时间序列分析在静安区流感样病例（ILI）预测预警中的应用。
方法 使用时间序列分析中的自回归求和移动平均模型（ARIMA）分别对静安区成人和儿童2011年7月-2014年7月的每月ILI监测发病数资料进行建模，然后预测2014年8-12月ILI发病水平，并与实际水平对比，以检验模型。
结果 监测结果显示，儿童和成人ILI发病趋势特点不同，儿童ILI发病未见明显季节性，成人ILI季节性较为明显。儿童监测数据识别最佳模型为ARIMA（0，1，1），成人监测数据识别最佳模型为ARIMA（1，0，0）（1，1，0）12。两个模型参数均具有统计学意义，预测效果良好。
结论 可使用ARIMA模型对静安区ILI发病趋势进行预测，但需对成人和儿童数据分别分析，根据预测区间可考虑设定预警阈值。
Abstract:Objective To apply time series analysis in forecasting and early warning of influenza-like illness (ILI) occurred in Jing'an District, Shanghai.
Methods The monthly monitoring data of ILI incidence of adults and children in Jing'an District during July 2011 and July 2014 were modeled with autoregressive integrated moving average model (ARIMA) in time series analysis. Then we predicted the ILI incidence from August to December 2014 by the established model, and tested the model by comparing predicting values and actual values.
Results The monitoring results showed different trends of ILI between children and adults. There were obvious seasonal variations for adults, but not for children. The optimal model was ARIMA (0, 1, 1) for children and ARIMA (1, 0, 0)(1, 1, 0)12 for adults. The parameters of the two models were both statistically significant, and the result of forecast was ideal.
Conclusion ARIMA model could be used to predict the incidence trend of ILI in Jing'an District, but the data of children and adults require separate analysis. Warning thresholds could be set according to the range of forecasting results.
下载:
