【摘要】 目的:探讨利用两种时间序列模型预测上海肺结核发病趋势的可行性,为制定上海地区肺结核预防控制策略提供科学依据。方法:收集上海市2007年1月—2018年6月传染病历史疫情资料,对2007年1月—2017年12月肺结核月发病人数分别采用自回归移动平均模型(ARIMA)乘积季节模型与指数平滑模型进行拟合,预测2018年1—6月的肺结核月发病人数,并与真实值进行比较。结果:ARIMA(0,1,1)×(1,1,2)12为乘积季节模型的最优模型,均方根误差(RMSE)为76.27,2018年1—6月预测值的相对误差和为0.402;运用指数平滑法构建的最优模型是Holt-Winters加法指数平滑,均方根误差(RMSE)为69.61,2018年1—6月预测值的相对误差和为0.292。结论:指数平滑模型拟合效果较好,预测精度更高,对上海市肺结核疫情的防控具有一定的指导意义。更多还原

【Abstract】 Objective:To explore the feasibility of two time series models for predicting tuberculosis epidemic in Shanghai,so as to provide a scientific reference for the prevention and control of tuberculosis in Shanghai. Methods:The epidemiological data of tuberculosis from Jan.2007 to Jun.2018 in Shanghai was collected. The monthly cases of tuberculosis from Jan.2007 to Dec.2017 was fitted by multiple seasonal ARIMA model and the exponential smoothing model. We predicted the monthly number of tuberculosis cases from Jan. to Jun. 2018 using the established models and comparcd the results to the real values. Results:ARIMA(0,1,1)×(1,1,2)12 was the optimal ARIMA model whose RMSE was 76.27 and the sum of the relative error from Jan. to Jun.2018 was 0.402;The optimal model constructed by exponential smoothing was Holt-Winters additive exponential smoothing model with RMSE of 69.61 and a sum of relative error of 0.292 from Jan. to Jun.2018. Conclusion:The exponential smoothing model could be fitted more effectively and had higher predictive accuracy. In short,it has important guiding significance for the prevcntion and control of tuberculosis in Shanghai.更多还原