Waterloo student conference abstract submission

Title: A Bayesian multinomial spatio-temporal for HIV risk group proportions.

Abstract: The Global AIDS Strategy 2021-2026 identifies adolescent girls and young women (AGYW) as a priority population for HIV prevention, and recommends differentiating intervention portfolios according to local HIV incidence, and individual risk behaviour. To help countries implement the strategy, we developed a Bayesian multinomial spatio-temporal model to estimate district-level HIV risk group proportions for AGYW across 13 high burden countries in sub-Saharan Africa. We used the multinomial-Poisson transformation to enable fast and accruate inference for the model via integrated nested Laplace approximations. Using our risk group proportions we estimated additional policy-relevant indicators including HIV prevalence and HIV incidence by risk group, as well as numbers of infections that could be averted using different intervention strategies. For more information, see athowes/multi-agyw.