PROJECT OVERVIEW

Despite global progress in combating HIV, the World Health Organization (WHO) estimates that about 1 in 7 infected individuals remain undiagnosed in 2024. Traditional testing strategies are facing diminishing returns, and funding constraints, such as recent cuts to USAID, have increased the urgency for more resource-efficient approaches to HIV testing.

RESEARCH OVERVIEW

In collaboration with WHO and the Gates Foundation, we developed an adaptive AI-based testing strategy that helps public health teams identify as many undiagnosed cases as possible given a fixed testing capacity. The method leverages on the structure of the underlying transmission network and recommends an optimal sequence of individuals to test. In simulation studies, our approach consistently found more positive cases than baseline strategies using the same budget.

To push our method towards real-world impact, we are currently preparing for a trial in South Africa and a retrospective evaluation in Tanzania with FHI360. We are also working on extensions of our method and improving other processes in the space of disease testing.

This project highlights how AI, when grounded in real-world constraints, can improve epidemic response and advance global health goals. To find out more about our techniques and algorithms, please have a look at our publications or reach out to us.

PROJECT PARTICIPANTS

(Sorted in alphabetical last name ordering)

PARTNER ORGANIZATIONS AND AFFILIATIONS OF PROJECT PARTICIPANTS

MEDIA

To be updated.

RELATED PUBLICATIONS

(* Indicates equal contribution)

Davin Choo*, Yuqi Pan*, Tonghan Wang, Milind Tambe, Alastair van Heerden, Cheryl Johnson. Adaptive Frontier Exploration on Graphs with Applications to Network-Based Disease Testing, Conference on Neural Information Processing Systems (NeurIPS), 2025.