University of Washington - Department of Statistics
Advisor: Adrian Raftery
Every two years, the Joint United Nations Programme on HIV/AIDS (UNAIDS) produces probabilistic estimates and projections of HIV prevalence rates for countries with generalized HIV/AIDS epidemics. To accomplish this task, Estimation and Projection Package (EPP) has been developed by using a generic epidemiological model and data from antenatal clinics and household surveys. It has been used by the UNAIDS Secretariat and national officials to estimate and project HIV epidemics.
Initially, the posterior distribution in EPP was approximated by sampling-importance-resampling (SIR), which is simple to implement and gives acceptable results for most countries (Alkema, Raftery, and Clark, 2007). For some countries, however, SIR is not computationally efficient because the posterior distribution tends to be concentrated around nonlinear ridges and can also be multimodal. We develop an Incremental Mixture Importance Sampling (IMIS) algorithm which substantially improves the sampling efficiency of SIR while retaining its essential simplicity and transparency (Raftery and Bao 2010).
While EPP 2009 produces plausible fits to a wide variety of observed patterns in surveillance data, a number of specific patterns are difficult to reproduce. We propose a modification of the EPP model, called the R-stochastic model, in which the infection rate is allowed to vary across years and is applied to the entire non-infected population (Bao and Raftery 2010).
In most countries outside of sub-Saharan Africa, HIV is largely concentrated in sub-populations whose behavior puts them at high risk for contracting and transmitting HIV, such as intravenous drug users, commercial sex workers and men who have sex with men. Estimating the size of these sub-populations is important for assessing overall HIV prevalence and designing effective interventions. We develop a Bayesian hierarchical model for estimating the sizes of local and national HIV at-risk populations (Bao, Raftery and Reddy 2010). It incorporates multiple commonly used data sources including mapping data, surveys, interventions, capture-recapture data, estimates or guesstimates from organizations, and expert opinion. The proposed model is used to estimate the numbers of intravenous drug users in Bangladesh and yields satisfactory results.