THE CHAIN MULTINOMIAL MODELS OF THE HIV EPIDEMIOLOGY IN HOMOSEXUAL POPULATIONS

In this paper, we propose a general class of stochastic models for the HIV epidemic in homosexual populations. This class of models are referred to as chain multinomial models. We illustrate this class of models by assuming some stochastic models of the HIV epidemic in homosexual populations in Sections 2-4 and derive the means, the variances and the covariances; we also derive results for the mean behaviors of these models. It is shown that under proportional mixing, the mean behaviors and the trend of these models are quite similar to one another; yet the variances and covariances of the infective people and AIDS cases of these models are very different over different models. These results suggest that the infection duration of infective people and sexual activity levels have significant impact on the projection of HIV epidemic.