Leveraging big data science to focus the HIV response in countries with generalized HIV epidemics - the overarching goal of the proposed aims is to leverage novel methods with large and…

Leveraging big data science to focus the HIV response in countries with generalized HIV epidemics - the overarching goal of the proposed aims is to leverage novel methods with large and underutilized data sets to evaluate the potential impact of increasingly specific HIV responses across generalized epidemic settings in sub-saharan africa (SSA) in reducing overall HIV incidence. This application is highly responsive to multiple areas of interest in the recent notice of special interest (NOSI): harnessing big data to halt HIV (not-AI-21- 054). Moreover, these aims align with current realities of the HIV pandemic. While overall incidence has steadily declined over the last 15 years, over 1.5 million people newly acquired HIV in 2020 including one million people across SSA. The risk for HIV is not evenly distributed anywhere in the world. And while specific key populations are recognized to be at increased risk of HIV in many higher income settings, a general population construct is often used to represent HIV epidemics across SSA. This construct typically negates proximal determinants of HIV acquisition and transmission, including heightened transmission risks in the contexts of condomless SEX between men, SEX work, and drug use, as well as infections among transgender people and incarcerated populations. We propose an ambitious set of aims that will leverage available HIV-related data for key populations as well as auxiliary data including from social media, search patterns, spatial data, socioeconomic and migration data. We will assemble multiple data sources and integrate these data to build a comprehensive data warehouse to estimate key population-specific indicators including HIV incidence and prevalence, population size, engagement in the HIV treatment cascade, and structural determinants. These estimates, augmented by small area estimation methods where data are sparse, will inform dynamic transmission models to estimate differential risks of onward HIV transmission among key populations and to better address the needs of key populations compared with general-population approaches. Finally, we will leverage very large and underutilized program data for HIV testing, prevention, and treatment programs in partnership with implementing partners. Cameroon, kenya, senegal, and south africa will be used as exemplar countries given that there exists sufficient data, willing governments, and they represent common HIV epidemic typologies in their respective regions of SSA. Aim 1: build a flexible, comprehensive, and accessible data warehouse collating available HIV-related and relevant auxiliary data for key populations from 2000 onward in SSA. Aim 2: employ small area estimation methods and spatial statistics using available direct and auxiliary data to infer population size, prevalence, and engagement in the treatment cascade for key populations. Aim 3: characterize the transmission population attributable fraction for HIV among key populations in each setting, incorporating differential risks of onward HIV transmission over multiple time horizons. Aim 4: evaluate routinely collected program data to inform tailoring and adaptation of implementation strategies for delivery of HIV prevention and treatment for key populations.