AI for HIV Impact: WHO Perspectives on Priorities and Challenges

Talk abstract.

Speaker bio. Dr. Cheryl Case Johnson has served as a Technical Officer in the Global HIV, TB, Hepatitis, and STIs programs at the World Health Organization (WHO) for over a decade. She is a member of the Lancet Commission on HIV and actively leads the innovation workstream which focuses on advancing strategic use of AI and digital health, self-care and integration.

AI for HIV Prevention: From Influence Maximization and Branching Bandits to Generative Scaling

For over two decades, my team and I have focused on advancing AI for social impact, addressing critical challenges in public health, conservation, and public safety. A central theme of our work is the optimization of limited intervention resources. Since 2014, we have specifically developed network-based AI interventions for HIV prevention in collaboration with diverse partners. In this talk, I will highlight two key projects in this domain. The first utilizes social network influence maximization to identify peer leaders for behavioral change interventions among marginalized populations. The second introduces a framework for network-based HIV testing that maps sparse transmission networks to branching bandits, leveraging Gittins indices to prioritize testing sequences under strict resource constraints. To bridge the "data gap," I will also discuss policy-embedded graph expansion using generative diffusion models to predict underlying transmission structures. While primarily focused on HIV, I will briefly highlight the expansion of these resource allocation methodologies to maternal health initiatives in India and the emerging potential for generative AI approaches, including LLM agents and diffusion models, to scale social good interventions.

Speaker bio. Milind Tambe is the Gordon McKay Professor of Computer Science at Harvard University; concurrently, he is Principal Scientist and Director for “AI for Social Good” at Google Research. Prof. Tambe and his team have developed innovative AI and multi-agent reasoning systems that have been successfully deployed to deliver real-world impact in public health (e.g., maternal and child health), public safety, and wildlife conservation. He is the recipient of the AAAI Award for Artificial Intelligence for the Benefit of Humanity, the AAAI Feigenbaum Prize, the IJCAI John McCarthy Award, the AAAI Robert S. Engelmore Memorial Lecture Award, and the AAMAS ACM/SIGAI Autonomous Agents Research Award. He is a fellow of AAAI and ACM. His contributions in Operations Research and public safety have also been recognized with the INFORMS Wagner Prize for excellence in Operations Research practice, Military Operations Research Society Rist Prize, the Columbus Fellowship Foundation Homeland security award, and commendations and certificates of appreciation from the US Coast Guard, the Federal Air Marshals Service, and airport police at the city of Los Angeles.

From Clinics to Code: Operationalizing AI for Health in Low- and Middle-Income Countries

Artificial intelligence in global health is often discussed in aspirational terms, yet implementation in low- and middle-income countries requires grappling with real constraints: scarce resources, fragmented data systems, stigma, workforce shortages, and the social and community dynamics that ultimately determine whether interventions succeed or fail. Drawing on a portfolio of public health research across sub-Saharan Africa and South Asia, this talk presents the messy, human reality of delivering health services at scale. Drawing on concrete examples spanning HIV prevention, mental health, diabetes, and tuberculosis the talk discusses the optimization of scarce resources across social networks, supporting overstretched community health workers, navigating stigma in sensitive health conversations, detecting risk early before conditions escalate, and understanding how communities perceive and trust the technologies being deployed among them. Each example surfaces a different class of data, a different operational decision, and a different kind of gap between what we can measure and what we need to act on.

Speaker bio. Alastair van Heerden is Research Director of the Syndicate for Public Science and Emerging Technologies (SYNAPSE). He has over 15 years of experience conducting clinical, behavioral and community-based research throughout East and Southern Africa, the United States, Nepal and Brazil. He has an interdisciplinary focus to his research which combines his interest in technology for development and public health with the aim of improving access to care for underserved and poorly resourced communities.

Deploying AI4SG Applications: Moving towards Sustainable Change

The AI for Social Good (AI4SG) community has produced an impressive portfolio of applications: predictive models for disease risk, optimization tools for resource allocation, decision-support systems for frontline workers, and reinforcement learning approaches for adaptive interventions. Many of these systems demonstrate strong technical performance and generate enthusiasm in pilot studies. Yet a sobering reality remains: a significant fraction do not achieve sustained real-world impact. In this talk, I contrast two categories of AI4SG projects. The first includes technically sound, well-published systems that succeed in controlled trials or short-term pilots but fail to persist beyond grant cycles or research partnerships. The second includes applications that have achieved durable integration into institutional workflows and demonstrable, ongoing impact. What differentiates them? In this talk, we will try to start a discussion on what differentiates the former from the latter. Is it primarily a question of technical design, or of institutional alignment? Do successful deployments hinge on choosing the “right” AI technique, or on embedding systems within existing decision-making structures? How much do sustainability, maintenance planning, incentive alignment, and local capacity matter relative to predictive accuracy or methodological novelty?

Speaker bio. Amulya Yadav is an associate professor in the College of Information Sciences and Technology at Penn State, where he directs the RAISE Lab @ Penn State. He also serves as the Associate Director of the Center for Socially Responsible AI @ Penn State. His research interests include artificial intelligence, multi-agent systems, computational game-theory, and applied machine learning. His work in the field of Artificial Intelligence for Social Good focuses on developing theoretically grounded approaches to real-world problems that can have an impact in the field. His algorithms have been deployed in the real world, particularly in the field of public health and wildlife protection. Amulya's leadership of the RAISE Lab has received an Exemplary Designation at the W.K. Kellogg Community Engagement Awards 2024. Amulya's papers have also received numerous awards, such as the PRICAI 2025 Best Student Paper Award, the AAMAS 2016 Best Student Paper Award, the AAAI 2017 Best Video and Best Student Video Award, the IDEAS 2016 Most Visionary Paper Award, and the AAMAS 2017 Best Paper Award nomination. His work has also been highlighted by Mashable.com as one of 26 incredible innovations that improved the world in 2015.

Combining Lenacapavir + AI to Drastically Reduce HIV Transmission

Lenacapavir (LEN), a twice-yearly injectable shown to be nearly 100% effective at preventing HIV, represents a transformative opportunity to achieve epidemic control, A CHAI-negotiated price reduction for Generic LEN from $28,000 to $40 per person per year has made large-scale deployment feasible, but reaching the populations who would benefit most—including adolescent girls, commercial sex workers, men who have sex with men, and transgender women,—remains the central implementation challenge. AI-driven targeting approaches, including geospatial micro-targeting at higher granularity than existing health information systems and predictive risk modeling in populations where stigma limits traditional identification methods, will be essential to closing this gap. Beyond targeting, AI offers additional high-impact opportunities across the rollout value chain, from demand forecasting and supply chain optimization to patient-facing chatbots and predictive models for treatment adherence and loss-to-follow-up. Realizing LEN's potential to avert over a million infections and billions in treatment costs will depend not only on funding and health system readiness, but on the intelligent, data-driven deployment strategies that AI can uniquely enable.

Speaker bio. Ben Brockman (CHAI Senior Director, AI): Ben leads CHAI’s Global AI team focused on supporting Ministries of Health in leveraging the most advanced analytics to improve health outcomes for patients. He spent the first decade of his career at IDinsight.org, where he was built a 20+ person AI-for-good consulting team. Ben also has experience in the private sector as GenAI lead for CVS Health's Specialty Pharmacy division and on BCG’s AI Go-to-Market team. Ben holds a Bachelors from the University of Pennsylvania and Masters from the Harvard Kennedy School and lives in Somerville, MA with his wife and daughter.
Speaker bio. Stephanie Dowling (CHAI Associate Director, Pediatric HIV): Stephanie has 9 years of experience supporting HIV and health systems programs across Sub-Saharan Africa, holding a Master of Public Health in health policy and global health. Stephanie has been with CHAI for 9 years and currently leads the pediatric HIV program, working with Ministries of Health to optimize pediatric and adolescent testing, treatment, and prevention programs. She is also currently supporting SSA and SEA countries with strategic and evidence-based HIV resource optimization to navigate foreign aid cuts, while also supporting community readiness for scale-up of LEN and other priority PrEP products

AI for Epidemiology

Fueled by increasing availability of data sources and various initiatives (and also the recent COVID pandemic), there has been an increasing interest in data-centric computational solutions for diseases and public health more broadly. However, AI and ML approaches still remain underexplored. In this talk, I will go over recent work from our group on developing data centric AI methods for challenging spatio-temporal and network-based epidemiological problems. We will also discuss their applications (e.g. COVID, Hospital Acquired Infections, Pediatric hospitalizations) and also describe how these techniques can lead to general improvements for tasks across multiple disciplines.

Speaker bio. B. Aditya Prakash is the Associate Chair for Academic Affairs and a Professor of Computing at the Georgia Institute of Technology (“Georgia Tech”). He received a Ph.D. from the Computer Science Department at Carnegie Mellon University, and a B.Tech in CSE from the Indian Institute of Technology (IIT) -- Bombay. He has published one book, more than 125 papers in major venues, holds three U.S. patents and has given several tutorials at leading conferences. His work has also received multiple best-of-conference, best paper and travel awards. His research interests include Data Science, Machine Learning and AI, with emphasis on big-data problems in large real-world networks and time-series, with applications to computational epidemiology/public health, urban computing, security and the Web. Tools developed by his group have been in use in many places including ORNL, Dow and Walmart. He has received several awards such as Facebook Faculty Awards, the Dolby Faculty Award, the NSF CAREER award, the IIT Bombay Young Alumni Achiever Award, and was named as one of ‘AI Ten to Watch’ by IEEE. He was also an invited participant at the NAE FOE series. His work has also won top prizes in multiple data science challenges (e.g the Catalyst COVID19 Symptom Challenge and the Harvard PET challenge) and been highlighted by several media outlets/popular press like FiveThirtyEight.com. He is also a member of the infectious diseases modeling MIDAS network and core-faculty at the Center for Machine Learning (ML@GT) and the Institute for Data Engineering and Science (IDEaS) at Georgia Tech. His webpage is: https://faculty.cc.gatech.edu/~badityap/

Rose's 'prevention paradox' and HIV

The availability new highly efficacious antiretroviral-based HIV prevention products have generated enthusiasm for efficiently deploying these products to accelerate reductions in new infections towards achieving global incidence reduction targets by 2030. However, the nature of contemporary HIV transmission dynamics in declining epidemics makes it hard to predict who will acquire HIV infection due to a majority of new infections arising among a large population with low- to moderate individual risk—creating a classic example of Geoffrey Rose's 'prevention paradox'. These challenges are compounded by limited contemporary data on who is acquiring HIV infection and missed opportunities to intervene. This talk will describe challenges and opportunities to address efficient deployment of HIV prevention in high HIV burden settings in southern and Eastern Africa.

Speaker bio. Jeff Imai-Eaton is an Associate Professor of Epidemiology and the Associate Director of the Center for Communicable Disease Dynamics at the Harvard T.H. Chan School of Public Health. His team's research involves developing mathematical and statistical models for characterizing HIV epidemic trends, transmission dynamics, and the impacts of HIV in Africa, including several tools supported by UNAIDS used for national HIV epidemic estimates.

Matching Machine Learning Capabilities to Public Health Needs

Recent progress in AI undoubtedly provides unprecedented opportunities for public health in general and HIV in particular, but between the sensational hype and the well-publicized catastrophes, it’s hard to know exactly how to take advantage of them. Neither the hype nor the catastrophes – well, most of them, anyway – are unique to AI. I will suggest some ways to think about appropriate uses and guardrails from a traditional Machine Learning perspective informed by the BioComplexity Institute’s experiences in modeling infectious disease epidemiology.

Speaker bio. Stephen Eubank is a professor at the BioComplexity Institute and in the Dept. of Public Health Sciences, both at the University of Virginia. His degrees are in physics and he has worked in the fields of fluid turbulence (La Jolla Institute); nonlinear dynamics and chaos (Los Alamos National Laboratory and the Santa Fe Institute); financial market modeling (co-founder, Prediction Company); ecological time series analysis (Biosphere 2); and natural language processing (Advanced Telecommunication Research in Kyoto). As a staff member at LANL, Dr. Eubank played leading roles in developing the Transportation Analysis and Simulation System (TRANSIMS); the Epidemiology Simulation System (EpiSims) project; and the Urban Infrastructure Suite (UIS). UIS is a collection of interoperable simulations of interacting infrastructures, each of which simulates the behavior of every individual in a large urban region. The goal of UIS is to model the dynamics of systems that include both physical and social components. In the BioComplexity Institutes, first at Virginia Tech and currently at the University of Virginia, he contributes to developing advanced technology for the study of large socio-technical systems and understanding micro/macro scaling in dynamical systems.