HIV is the largest single cause of death among adults in Sub-Saharan Africa, responsible for about a third of all deaths among adults. One of the key paradigms to halting HIV in Sub-Saharan Africa relies on identification of infected individuals and populations for delivery of biomedical and behavioral interventions. However, by the end of 2015 less that half of HIV-infected individuals accessed antiretroviral therapy (ART) despite expansion of eligibility and ongoing efforts to diagnose and initiate treatment. A better understanding of the social, behavioral, environmental, and economic contexts that influence HIV risk could improve the effectiveness and efficiency of programs that aim to identify and target HIV-infected populations. In response to the program announcement for ?Harnessing Big Data to Halt HIV? (PA-15-273), the overall goal of this proposal is to develop new analytic tools in large-scale data to predict risk of HIV infection and to generate hypotheses about new or under-recognized risk factors in Sub-Saharan Africa. We plan four primary investigations: (1) Extract and harmonize all Sub- Saharan African nationally representative Demographic and Health Surveys that include the HIV status of over 600,000 men and women collected in 29 countries, and hundreds to thousands of associated exposure variables; (2) Develop analytic tools based on LASSO and XWAS to predict HIV infection status and generate hypotheses about social, behaviorial, environmental, an economic risk factors; (3) Identify HIV risk in multi-country, large-scale data and synthesize findings across in Sub-Saharan Africa, and (4) develop a bioethics program to identify targets for new interventions and policies in a culturally and ethically sound manner. The project will develop leverage big data and high-throughput analytic methodology in the service of global HIV control. The outputs of this project include (1) accessible software code for efficient exploration of robust correlates of HIV status derived from the biggest collection of high-dimensional, harmonized, and nationally representative representative household surveys in the world, (2) an extensive landscape of the social, environmental, behavioral, and economic factors predictive of HIV infection among over 600K people tested for HIV in 29 Sub- Saharan countries, and (3) a ethical framework to enable practical, relevant, and appropriate translation and communication of findings. New models of HIV infection will facilitate identification of at-risk groups and the development of interventions to halt the HIV epidemic in Sub-Saharan Africa.

Public Health Relevance

Advancing the the public health goals for global HIV, including the ?AIDS-Free Generation? and ?90-90- 90,? requires effective identification of HIV-infected individuals. The overall goal of this proposal is to develop new analytic tools to (1) predict HIV status using modern tools of statistical machine learning and (2) uncover new and under-recognized risk factors for HIV infection in Sub-Saharan Africa. Using person-level data including the HIV status of over 600,000 men and women collected in nationally representative household surveys from 29 countries, we propose to build a high-throughput statistical predictor of HIV status and apply a novel ?X-wide association study? (XWAS) to search for and validate social, behavioral, environmental, and economic factors associated with HIV seropositivity. Through development of an ethical/decision making framework and with close consult of an advisory committee, the findings will be used to develop both broad and specific insights about risk factors and at-risk populations for improving HIV care and epidemiology.

Agency
National Institute of Health (NIH)
Institute
National Institute of Allergy and Infectious Diseases (NIAID)
Type
Research Project (R01)
Project #
5R01AI127250-02
Application #
9460355
Study Section
Special Emphasis Panel (ZRG1)
Program Officer
Mckaig, Rosemary G
Project Start
2017-04-01
Project End
2021-03-31
Budget Start
2018-04-01
Budget End
2019-03-31
Support Year
2
Fiscal Year
2018
Total Cost
Indirect Cost
Name
Stanford University
Department
Internal Medicine/Medicine
Type
Schools of Medicine
DUNS #
009214214
City
Stanford
State
CA
Country
United States
Zip Code
94304
Patel, Chirag J; Kerr, Jacqueline; Thomas, Duncan C et al. (2017) Opportunities and Challenges for Environmental Exposure Assessment in Population-Based Studies. Cancer Epidemiol Biomarkers Prev 26:1370-1380