ARIMA模型在北京市丰台区流行性感冒预测中的应用

Application of ARIMA model in forecasting incidence of influenza in Fengtai District of Beijing

目的探讨ARIMA模型在北京市丰台区流行性感冒预警预测中的应用,建立流感样病例(ILI)发病预测模型。方法利用《北京市医院传染病监测预警信息系统》资料,采用SPSS 22.0软件,对丰台区2007年9月—2016年8月ILI监测数据建立ARIMA模型,并对2016年9月—2017年4月ILI%进行预测,评价模型的预测效果。结果 2007年9月—2016年8月丰台区ILI%呈季节性周期变化,病例以<25岁儿童及青少年为主,主要就诊科室为儿科门、急诊和发热门诊。2007年9月—2016年8月ILI%数据拟合模型ARIMA(1,0,0,)(1,1,0)12,预测2016年9月—2017年4月ILI%与监测值差异无统计学意义(t=1.674,P>0.05)。结论 ARIMA模型可用于丰台区流行性感冒发病情况预测,可为流行性感冒防控提供科学依据。

[Objective]To explore the application of auto-regression integrated moving average（ARIMA） model in prediction and early warning of influenza in Fengtai District of Beijing,establish a prediction model for influenza like illness（ILI）.[Methods]According to the data of Beijing monitoring and early warning system for infectious diseases, the ARIMA model was established on monitoring data of ILI in Fengtai District from September 2007 to August 2016 by SPSS22.0. The ILI% from September 2016 to April 2017 was predicted to evaluate the prediction effect of the model.[Results]The monthly ILI% from September 2007 to August 2016 presented annual seasonality. The majority of ILI cases were children under 5 years old and adolescents. The clinic departments were mainly fever outpatient service, pediatric outpatient and emergency department. The optimal model to ILI rate from September 2007 to August 2016 was ARIMA（1,0,0,）（1,1,0）12. No statistical difference（t=1.674,P 0.05） was found between prediction and monitoring values of ILI% from September 2016 to April 2017.[Conclusion]The ARIMA model can be applied to forecasting and early warning of influenza in Fengtai District, providing scientific evidence for prevention and control of influenza.