Forecasting the AIDS epidemic in Puerto Rico

The purpose of this study is to model and predict the diffusion of the AIDS epidemic in Puerto Rico. Specifically we aim at identifying primary influences in the geographical distribution of the population affected with AIDS to produce a reasonable projection of the time and space paths that will be followed by the epidemic. The study is based on AIDS incidence data from 1982 through 1992. The epidemic is analyzed statistically through multivariate regression and the potential model to produce a probability surface in which risk behavior and intra-municipality mobility are significant risk factors. The growth forecast is produced using the logistic function fitted to the past growth using an iterative non-linear optimization method. A simulation technique is then employed to forecast the spatial development of the epidemic from 1993 through 1997. The results indicate a clear hierarchical tendency at the beginning of the epidemic, later a wave-like diffusion pattern is also observed. While the absolute number of new cases is expected to remain higher in the more urbanized areas, the relative growth of AIDS cases is likely to become much higher in the rural municipalities. The forecasting procedure employed here is applicable to populations with diverse epidemiological profiles.