A Joint Spatial Conditional Auto-Regressive Model for Estimating HIV Prevalence Rates Among Key Populations

TL;DR

This paper introduces a joint spatial conditional auto-regressive model that improves the estimation of HIV prevalence among key populations by leveraging spatial and population dependencies, aiding policy decisions.

Contribution

It presents a novel joint spatial model that borrows information across locations and populations, enhancing accuracy over traditional independent methods.

Findings

The model provides more accurate HIV prevalence estimates.

It outperforms independent sub-epidemic fitting in real data.

Theoretical and numerical studies reveal conditions for better predictions.

Abstract

Ending the HIV/AIDS pandemic is among the Sustainable Development Goals for the next decade. In order to overcome the gap between the need for care and the available resources, better understanding of HIV epidemics is needed to guide policy decisions, especially for key populations that are at higher risk for HIV infection. Accurate HIV epidemic estimates for key populations have been difficult to obtain because their HIV surveillance data is very limited. In this paper, we propose a so-called joint spatial conditional auto-regressive model for estimating HIV prevalence rates among key populations. Our model borrows information from both neighboring locations and dependent populations. As illustrated in the real data analysis, it provides more accurate estimates than independently fitting the sub-epidemic for each key population. In addition, we provide a study to reveal the conditions that our proposal gives a better prediction. The study combines both theoretical investigation and numerical study, revealing strength and limitations of our proposal.