基于向量自回归模型的清远市手足口病发病预测分析

目的运用向量自回归模型(vector autoregression model,VAR)分析清远市手足口病发病与气象因素的动态关系,为手足口病的防控提供科学依据。方法用2013年1月1日-2017年4月30日的气象数据和手足口病发病数据建立VAR模型,评价模型的稳定性及拟合效果,通过脉冲响应函数和方差分解分析对模型进行定量分析,进一步对手足口病发病情况进行预测并评价预测效果。结果模型总拟合优度为0. 96,调整后拟合优度为0. 95。手足口病发病数对日最低气温、日地表最低气温和日平均相对湿度等气象因素的冲击响应是正向的。模型对未来1周、2周、3周和4周的手足口病发病预测其平均绝对百分误差(mean absolute percentage error,MAPE)分别为17. 75%、13. 69%、10. 74%和15. 85%,希尔不等式系数(Theil inequality coefficient,TIC)分别为0. 08、0. 06、0. 05和0. 09。结论气象因素和手足口病发病的VAR模型可以较好地进行清远市手足口病发病的短期预测。

Prediction of hand,foot and mouth disease incidence in Qingyuan City based on vector autoregression model

Objective To analyze the dynamic relationship between hand, foot and mouth disease (HFMD) incidence and meteorological factors in Qingyuan City by using vector autoregression model (VAR) so as to provide a scientific basis for prevention and control of HFMD. Methods The VAR model was established with the meteorological data and HFMD incidence data from January 1, 2013 to April 30, 2017. The stability and fitting effect of the model were evaluated. The model was analyzed quantitatively by using impulse response function and variance decomposition analysis, and the results were used to further predict the incidence of HFMD and evaluate the prediction effect. Results The total goodness-of-fit of the model was 0.96, and the adjusted goodness-of-fit was 0.95. The impact of HFMD incidence on the daily minimum temperature, daily minimum surface temperature and daily average relative humidity was positive. The mean absolute percentage errors in the prediction of HFMD in the next week and next 2, 3 and 4 weeks were 17.75%, 13.69%, 10.74% and 15.85% respectively, and Theil inequality coefficients were 0.08, 0.06, 0.05 and 0.09 respectively. Conclusions The VAR model of meteorological factors and HFMD incidence can be used for short-term prediction of HFMD incidence in Qingyuan City.