ARIMA乘积季节模型在江西省手足口病发病预测中的应用

目的 建立江西省手足口病发病的ARIMA乘积季节模型并进行预测,探讨该模型在江西省手足口病早期预测预警中的可行性。方法 以江西省2008-2015年手足口病逐月发病数为基础,建立ARIMA乘积季节模型,利用2016年各月发病数对模型进行验证,得到最优模型后对2017年手足口病发病数进行预测。结果 本研究构建的江西省手足口病发病预测模型为ARIMA(2,1,1)(0,1,1)12,各项参数都有统计学意义(P<0.05),模型拟合优度检验获得BIC值最小为14.738,残序列为白噪声(Ljung-Box Q=14.738,P=0.362),实际值与拟合值的动态趋势基本吻合;2016年1-8月预测值与实际值相差较小,吻合度高;2017年手足口发病数预测值为49 353。结论 ARIMA(2,1,1)(0,1,1)12较好地模拟了江西省手足口病发病趋势,可用于江西省手足口病发病的短期预测。

Application of ARIMA product seasonal model in the incidence prediction of hand foot and mouth disease,Jiangxi

Objective To build an ARIMA product seasonal model of hand foot mouth disease(HFMD),and to explore the feasibility of the model in the early warning and incidence prediction of hand foot mouth disease in Jiangxi province. Methods The ARIMA product seasonal model was built based on the monthly incidence cases from 2008 to 2015 in Jiangxi province and verified by using the data in 2016.The optimal model after verification was applied to predict the incidence of HFMD in 2017. Results The prediction model of HFMD was ARIMA(2,1,1)(0,1,1)12,and the parameters were all statistically significant(P<0.05).The minimum model goodness of fit value of BIC was 14.738,and the residual sequence was white noise(Ljung-BoxQ=14.738,P=0.362).The dynamic trends of actual values and fitted values were basically consistent.The difference between actual values and fitted values from January to August in 2016 was small and the degree of anastomosis was high.The predicted value of HFMD in 2017 was 49353. Conclusion The model can well simulate the trend of HFMD,and it can be used for the short-term prediction of HFMD in Jiangxi Province.