ARIMA模型在手足口病预测预警中的应用

目的研究时间序列分析在手足口病预测预警中的应用,并探讨提高模型预测准确性和实用性的思路。方法应用SPSS 18.0软件对深圳市盐田区2008年1月—2014年4月手足口病发病率进行ARIMA模型拟合,预测2014年5月-12月手足口病发病率。结果模型ARIMA（0,1,0）（0,1,1）12的参数估计值为0.761,t=2.552,P=0.013,经检验参数有统计学意义。且BIC=6.066,在拟合比较的模型中最小,故选定为最佳拟合模型。利用2013年10月—2014年4月实际发病数与预测发病数进行比较,实际值与预测值相对误差的中位数为71%。结论用时间序列分析对手足口病发病情况的拟合结果满意,预测和预警效果良好。

The applied research of ARIMA model in forcasting and early warning of Hand-Foot-Mouth disease

Objective To make research into the application of ARIMA model in the forcasting and early warning the HandFoot-mouth disease（HFMD） and elaborate on the idea of improving the accuracy and the applicability of the model. Methods Using SPSS18.0 software in January to 2008 in Yantian District of Shenzhen city in April-2014 years of hand foot and mouth disease incidence rate of ARIMA model, forecast in 2014 May-12 months of hand foot and mouth disease incidence. Results The model of ARIMA（0, 1, 0）（0,1,1） 12 parameter estimation value is 0.761, t=2.552, P=0.013, the test parameters have statistical significance. And BIC=6.066, the smallest in the fitted model, so it is selected as the best fit model. Using 2013.10-2014.4 on the actual incidence and the predictive incidence were compared, the actual and predicted values of median relative error is 71%.Conclusion Using time series analysis results fit the incidence of hand foot and mouth disease with good results, forecasting and early warning.