河北省麻疹疫情时间序列的预测和预警分析  

Forecasting and warning measles by time-series analysis in Hebei

作　　者：刘曙光[1] 王立芹[2] 刘岩[1] 曹秀芬[1] 赵保刚[1] 朱晓敏[2] 

机构地区：[1]河北省疾病预防控制中心,河北石家庄050021 [2]河北医科大学公共卫生学院,河北石家庄050017

出　　处：《中国卫生检验杂志》2015年第17期2954-2956,共3页Chinese Journal of Health Laboratory Technology

基　　金：河北省卫生和计划生育委员会(ZL20140243)

摘　　要：目的对河北省麻疹疫情进行时间序列分析,评估当前和历史疫情,并对未来疫情进行预测预警,为制定控制麻疹疫情策略与措施提供新的科学依据。方法利用EViews 8.0对河北省2001年1月-2014年10月麻疹月发病数建立季节自回归滑动平均混合(SARIMA)模型,首先采用取对数、差分等方法对序列进行平稳化,然后进行模型参数的估计、检验,最优模型的筛选,最后进行预测分析。结果最终通过检验的最优模型是SARIMA(0,1,0)(3,1,2)12,表达式为(1+0.66B12+0.18B36)d[ln(mt+1),1,12]=(1+0.87B24)εt;Theil不等式系数=0.13,BP≈0,VP=0.02,CVP=0.98,模型拟合和预测良好。实际值均在预测值的95%可信区间,2014年12月的预测发病数呈下降趋势。结论 SARIMA模型适用于河北省麻疹疫情的短期预测分析,可以即时地评价现行控制措施和预警未来疫情。Objective To analyze the measles＇ epidemic by time- series analysis in Hebei,then to evaluate the past and present epidemic,to forecast and warn the future trend in order to provide the new scientific evidence to formulate the further strategies and measures that are preventing and controlling the disease. Methods Using EViews 8. 0 to establish a seasonal autoregressive moving average（ SARIMA） model for the incidence of measles in Hebei from January 2001 to October 2014 in Hubei,first to make the data sequence placid by the logarithmic transformation and difference method,then to estimate and test the parameters of model,meantime,to select an optimal model,finally to forecast and analyze. Results The selected optimal model was SARIMA（ 0,1,0）（ 3,1,2）12,and the equation was（ 1 ＋ 0. 66B12＋ 0. 18B36） d[（ ln（ mt＋ 1）,1,12） ]=（ 1 ＋ 0. 87B24） εt.Theil inequality coefficient was 0. 13,and BP≈0,VP = 0. 02,CVP = 0. 98,the fitting and forecasting was very well. All the actual values was falling in the 95% confident rang of the forecasted cases. The forecasted case of December 2014 was decreasing.Conclusion SARIMA model is suitable for measles short- term forecasting,and it can be used for evaluating the current measures and warning the future epidemic.

关 键 词：麻疹 时间序列 自回归滑动平均混合模型 预测 

分 类 号：R181.8[医药卫生—流行病学]