AI & HIV: Can Machine Learning Deliver Better Outcomes? | aidsmap

The promise of artificial intelligence (AI) to revolutionize healthcare is generating considerable excitement, but also a healthy dose of skepticism. At the recent Conference on Retroviruses and Opportunistic Infections (CROI) 2026, held February 22-26 in Denver, Colorado, scientists explored the potential of machine learning and generative AI to improve outcomes for people living with HIV. However, the path from impressive laboratory demonstrations to tangible benefits for patients remains unclear.

The Hype and the Hesitation

The discussion at CROI 2026, as reported by aidsmap, centered on whether AI represents a genuine shortcut to better HIV management, or a potential “short-circuit” – a promising technology that fails to deliver on its initial expectations. Dr. Ravi Goyal, an assistant professor at the University of California, San Diego, and a moderator for one of the AI sessions, voiced a common concern. He acknowledged the impressive capabilities of machine learning and generative AI in controlled settings, but questioned whether these translate into improved patient outcomes in the real world. “We’ve been told that it’s going to revolutionise public health, it’s going to revolutionise our healthcare system,” Dr. Goyal stated, “But if you’re like me, I don’t know, maybe you don’t quite believe the hype, maybe you haven’t quite seen it yet.”

This sentiment highlights a crucial point: the distinction between technological potential and practical application. While AI algorithms can analyze vast datasets and identify patterns that might be missed by human researchers, turning those insights into effective interventions requires careful validation and implementation.

What AI Could Bring to HIV Care

Despite the cautious outlook, the potential applications of AI in HIV care are numerous. Machine learning algorithms could, for example, be used to predict which individuals are at highest risk of treatment failure, allowing clinicians to tailor interventions accordingly. AI could also accelerate the development of new drugs and therapies by identifying promising drug candidates and optimizing clinical trial designs. Generative AI, a branch of AI focused on creating new content, might even assist in the development of personalized prevention strategies.

AVAC, a leading HIV prevention research advocacy group, highlighted at CROI 2026 the importance of tracking advances in long-acting HIV prevention, evolving science around a cure, and the impact of global funding crises on research. They provided infographics on HIV R&D and prevention options to aid in discussions at the conference and beyond.

The Challenges of Implementation

Several challenges stand in the way of realizing AI’s full potential in HIV care. One major hurdle is the availability of high-quality data. Machine learning algorithms require large, well-curated datasets to train effectively. Data privacy concerns and the lack of interoperability between different healthcare systems can make it difficult to access and share the necessary data.

Another challenge is the potential for bias in AI algorithms. If the data used to train an algorithm reflects existing disparities in healthcare access or treatment, the algorithm may perpetuate those disparities. Ensuring fairness and equity in AI-driven healthcare requires careful attention to data collection and algorithm design.

Understanding Machine Learning and Generative AI

For those unfamiliar with the terms, machine learning is a type of artificial intelligence that allows computers to learn from data without being explicitly programmed. Algorithms are trained on datasets and then used to make predictions or decisions. Generative AI takes this a step further, creating new data that resembles the data it was trained on. In the context of HIV, this could mean generating new protein structures for potential drug targets or creating personalized risk assessments.

The Role of Biostatistics and Network Science

Experts like Dr. Ravi Goyal, whose work focuses on biostatistics, network science, and modeling complex systems (University of California, San Diego), emphasize the importance of rigorous statistical methods in evaluating the performance of AI algorithms. Network science, in particular, can be used to understand how HIV spreads through populations and to identify key individuals to target with prevention efforts. However, even sophisticated statistical models cannot fully account for the complexities of human behavior and social determinants of health.

What’s Next for AI and HIV?

The conversation surrounding AI and HIV is far from over. Researchers are continuing to explore new applications of these technologies, and ongoing clinical trials will help to determine whether AI-driven interventions can truly improve patient outcomes. The annual Conference on Retroviruses and Opportunistic Infections (CROI) remains a crucial venue for sharing research findings and fostering collaboration in this rapidly evolving field. AVAC will continue to track news and developments from the conference, and is offering virtual webinars called Community Breakfast Clubs to discuss the implications of the science presented at CROI.

Looking ahead, a key focus will be on developing robust methods for validating AI algorithms and ensuring their responsible implementation. This will require collaboration between researchers, clinicians, policymakers, and community advocates. It will also necessitate a commitment to data privacy, equity, and transparency. The ultimate goal is to harness the power of AI to improve the lives of people living with and at risk of HIV, but only if we proceed with caution and a clear understanding of both the opportunities and the limitations.