一种用于推算感染时间曲线的非参数极大似然估计方法

A non-parametric maximum likelihood estimation method for estimating infection curve of infectious diseases

目的:研究通过传染病发病时间曲线推算感染时间曲线的方法。方法:采用基于EMS算法的非参数极大似然估计方法,根据传染病的发病资料和潜伏期资料推算感染时间曲线,并用于中国内地SARS感染时间曲线推算。结果:通过Mat-lab编程,所采用的方法可以方便地估计出感染时间曲线,并可平滑个体数据的波动。用于推算中国内地SARS感染时间曲线,发现感染时间曲线平滑稳定,有明显的2个高峰,感染高峰与实际发病高峰间隅5 d左右;发病人数(4 634例)与估计的感染人数(4 633.6例)相差在0.5例以内。结论:基于EMS算法的非参数极大似然估计3-法对于感染时间曲线的估计是可靠的,可以用于SARS感染时间曲线的推算。所推算的感染时间曲线有助于干预措施被果评价。

Objective:To develop a method for constructing the infection curve with the incidence curve of infectious diseases. Methods: The incidence and incubation period data of infectious diseases were used to estimate the infection curve by a non-parametric maximum likelihood estimation (MLE) method based on EMS algorithm, which was used to construct the infection curve of severe acute respiratory syndrome(SARS) in China's Mainland. Results: The results showed that the programming method by software Matlab could easily construct the infection curve and smooth the individual fluctuations. When used for estimation of the infection curve of SARS in China's Mainland,it was found that the curve was smooth and had 2 obvious peaks. There were about 5 d interval from the peak of infection curve to that of incidence curve. The difference between incidence number (4 634) and estimated infection number (4 633. 6) was under 0. 5. Conclusion: The non-parametric MLE method based on EMS algorithm is reliable for the estimation of the infection curve and can be used to construct the infection curve of SARS,which will help us to evaluate the effectiveness of intervention measures.