传染病发病数资料的EM估计

E-M Algorithm for Infectious Disease Data on Outbreak Size

目的:旨在探讨EM算法在具有链结构的传染病发病数资料分析中的应用。方法:借助EM算法,对4口之家的麻疹发病数资料分别拟合Greenwood链二项分布模型和Reed-Frost链二项分布模型。模型拟合效果的比较采用Pearsonχ2检验。结果:基于Greenwood链二项分布模型时,家庭内麻疹感染率为29.08%;而基于Reed-Frost链二项分布模型时,家庭内麻疹感染率为34.67%。Reed-Frost链二项分布模型的拟合效果优于Greenwood链二项分布模型。结论:采用EM算法处理和分析传染病发病数资料更简便易行。

Objective： To explore the use of E-M algorithm for the analysis of chain infectious disease data on outbreak size. Methods： With the E-M algorithm, the Greenwood and Reed-Frost chain binomial models are fitted for infectious disease data on measles outbreak size in household of size four. Results： Based on Greenwood chain binomial model, the infection percent is 29.08 % . Based on Reed-Frost chain binomial mod- el, the infection rate is 29.08 %. The Reed-Frost chain binomial model fitted better than the Greenwood chain binomial model. Conclusions： E-M algorithm can be employed more flexibly to deal with the infectious disease data on outbreak size.