基于专家建模器的ARIMA季节模型在手足口病预测预警中的应用

Application of ARIMA seasonal model in prediction and alert of hand-foot-mouth disease based expert modeler

目的运用专家建模器,探讨自动建立ARIMA季节模型在手足口病发病人数预测预警中的适用性。方法基于岚山区2008～2016年手足口病季度报告发病人数资料,运用SPSS20.0专家建模器建立ARIMA季节模型,2017年发病报告数据验证模型预测效果,最后预测2018年的发病人数。结果专家建模器自动建立的最优模型为ARIMA(1,0,0)(0,1,1)4,各参数均有统计学意义,模型最小BIC指标值为9.29,残差序列经检验为白噪声(Ljing-Box统计量Q=10.27,P=0.85,),拟合值基本接近实际值;2017年各季度预测值与实际值动态趋势基本一致,预测的平均绝对百分比误差9.0,平均绝对误差11.73,均方根误差31.49;预测2018年的手足口病发病人数为592人。结论运用专家建模器自动建立ARIMA季节模型,方法简单、高效,适合基层进行手足口病的预测预警。

Objective To explore the applicability of ARIMA seasonal model automatically established by expert modeler in the prediction and alert of hand-foot-mouth disease. Methods The ARIMA seasonal model was established by SPSS20.0 expert modeler based on the quarterly reported cases of hand-foot-mouth disease in Lanshan district from 2008 to2016,the data of the hand-foot-mouth disease quarterly reported cases of 2017 were served as the examined samples of the ARIMA seasonal model to prove the effect of prediction,then it of 2018 was predicted by the model. Results The best model automatically given by the expert modeler was ARIMA(1,0,0)(0,1,1)4.All hypothesis tests of parameter estimations had statistical significance.The minimum BIC index value of the model was 9.29,and the residual sequence was white noise(Ljing-Box Q=10.27,P =0.85).The fitted values were basically close to the actual values from 2008 to2016.The dynamic trends of predicted and actual values in 2017 were basically consistent.The mean absolute percentage error was 9.0,the mean absolute error was 11.73,and the square root error was 31.49.The estimated number of cases of hand-foot-mouth disease in 2018 was 592. Conclusion Establishing ARIMA seasonal model automatically by expert modeler is simple,efficient and suitable for centers for disease control and prevention at grassroots level to predict handfoot-mouth disease and give early warning.