自回归移动平均模型在北京市朝阳区手足口病发病预测中的应用

摘要

目的 通过构建时间序列自回归移动平均模型(autoregressive integrated moving average model,ARIMA),对手足口发病趋势进行预测,探讨该模型在发病预测中的应用.方法 从疾病监测信息报告管理系统提取北京市朝阳区2010年1月-2016年12月手足口病月发病数据.建立ARIMA季节乘积模型,对2010年1月-2015年12月的月发病数进行拟合,再以2016年1-12月的月发病数作为验证数据,评价其预测效果.结果 通过对模型进行拟合优度及残差序列进行白噪声检验,最后选择了ARIMA(1,0,0)(1,1,0)12为最佳模型.对2016年1-12月发病数进行预测,实际发病数均落入95％CI内,平均相对误差为49.37％.模型中加入2016年1-6月的月实际发病数,预测2016年7-12月的月发病数,平均相对误差为18.12％.结论 ARIMA季节模型可应用于手足口病等具有季节性变动特征的传染病预测.ARIMA模型短期预测手足口病的发病情况精度更高,可通过不断纳入新的实际观测值开展动态分析.ARIMA模型仅为一种数学工具,在实际防控及监测工作中,需要结合专业理论知识及具体情况进行分析.%Objective To predict incidence trend of hand-foot-and-mouth disease (HFMD) by using autoregressive integrated moving average (ARIMA) model and provide evidence for the prevention and control of HFMD. Methods A time series analysis model was established basing on the monthly incident cases of HFMD from 2010 to 2015 in Chaoyang of Beijing. The model was used to predict incidence of HFMD during January to June in 2016. Results ARIMA(1,0,0) (1,1,0) 12 was selected as the best model after white noise test of goodness of fit and residual sequence. The average of the relative error between actual and predicted values from January to December in 2016 is 49. 37％. The average relative error of the predicted incident cases from June to December in 2016 is 18. 12％. Conclusion ARIMA season model is suitable to predict incidence of HFMD. The ARIMA model is of higher accuracy in a short period than in a long one,which can be used to improve the prediction accuracy by continuously recruiting new observations. In actual practice of disease control,it is necessary to combine the professional knowledge with ARIMA model.