Probabilistic projections of HIV prevalence using Bayesian melding.

摘要: The Joint United Nations Programme on HIV/AIDS (UNAIDS) has developed the Estimation and Projection Package (EPP) for making national estimates short term projections of HIV prevalence based observed trends in antenatal clinics. Understanding uncertainty its related quantities is important more informed policy decision making. We propose using Bayesian melding to assess around EPP predictions. Prevalence data as well information input parameters model are used derive probabilistic - a probability distribution set future trajectories. relate clinic population account variability between clinics random effects model. Predictive intervals derived checking discuss predictions given by results procedure Uganda where peaked at 28% 1990; 95% prediction interval 2010 ranges from 1% 7%. (authors)