摘要
目的 探讨北京SARS发病的动态变化规律。方法 建立北京市SARS发病长期预测的中位数回归模型以及短期预测的自回归模型。结果 以 4月 2 4日～ 5月 11日的数据建模 ,对SARS进行长期预测的中位数回归模型为 :^Yi0 5=12 9 2 -4 7ti,预测其新增临床诊断病例数将于 5月 2 2日降至零病例 ;同时考虑时间 (ti)、前一天的新增临床疑似病例数 (Zi-1) ,对SARS进行短期预测的二阶自回归模型 (第一步 )为 :^Yi=79 95 2 6-0 2 773Yi-1+0 3 5 82Yi-2 +0 2 848Zi-1-2 8175ti。预测其新增临床诊断病例数将于 5月 2 1日降至零病例。结论 SARS发病长期预测用中位数回归较一般线性回归稳健 ;SARS发病短期预测的自回归模型取二阶较适宜。长期预测与短期预测中新增临床诊断病例数降至零病例的时间基本一致。在传染病的发病预测中使用中位数回归模型及自回归模型的方法具有广阔的应用前景。
Objective To study the dynamic laws of SARS in Beijing.Methods The median-regressive model and the autoregressive model of SARS were set up.Results Based on the data from Apr. 24 to May 11, the median-regressive model was set up to predict the long-term trend,and the autoregressive model of SARS to predict step wisely. According to the two models, the number of new cases would decrease to zero till May 22 and May 21,respectively.Conclusion The median-regressive model was robust for long-term prediction and the second order autoregressive model was fit for step-wise prediction. The date decreasing to zero in the thee predictions of the two models were near. Great prospect will see the combining application of the median-regressive model and the autoregressive model to predict the infectious disease.
出处
《中国卫生统计》
CSCD
北大核心
2004年第4期219-221,共3页
Chinese Journal of Health Statistics
基金
江苏省社会发展基金资助项目 (编号 :BS2 0 0 0 0 4 3)
