In an era when so many view artificial intelligence (AI) as a tool, virologists are sizing it up to potentially predict the next pandemic. AI shows potential in helping global health experts detect emerging infectious diseases earlier. Still, researchers stress that the tool must complement — not replace — traditional surveillance led by a multidisciplinary team.

That message is central in a recent research comment, which posits that AI, when paired with a One Health approach, could strengthen pandemic preparedness by identifying areas around the world in need of greater surveillance. The One Health framework integrates human, animal, and environmental health data to prevent, detect, and respond to emerging health threats, according to the World Health Organization.

Enhancing Risk Assessment

Recent outbreaks, including SARS-CoV-2, mpox, and avian influenza, underscore the growing challenges of controlling emerging infectious diseases. Climate change, animal production practices, land-use changes, and human encroachment into natural habitats continue to elevate the risk for pathogen spillover from animals to humans, the authors noted.

For pandemic preparedness, “an area of considerable research is addressing whether we can predict hotspots of disease emergence resulting from spillovers of viruses from animals into humans,” said Marion Koopmans, PhD, a virologist at Erasmus MC in Rotterdam, Netherlands, and a co-author of the comment.

The idea, she says, is to predict pandemics by integrating multiple data streams into AI models, including historic pandemic patterns, biological processes that cause spillovers, and drivers of disease emergence.

Drivers such as animal density, habitat type, climate conditions, human population density, and ecosystem disturbance can influence outbreak risk, Koopmans explained, but surveillance with these factors is currently an inexact tool. “While this is a long shot, there are examples where combining some of these factors may actually help to target where we should enhance our surveillance,” said Koopmans.

Once such hot spots are identified, metagenomic sequencing could detect pathogens in environmental samples, including wastewater, air, food, and soil. In this way, AI could help flag early signals that warrant closer investigation by human experts.

AI may also aid early risk assessment once a potential pathogen is detected, as viral genomes encode information related to transmissibility and disease severity. “AI-based protein models can provide insight into what a mutation does to the structure of viruses, and how that may translate into the risk for spread or severe disease,” Koopmans said. Although the field remains technically challenging, she added, “we see great potential for the use of AI to speed up risk assessment.”

Raising Questions of Quality Control

More broadly, AI systems are being explored as “co-scientists” capable of synthesizing literature, generating hypotheses, analyzing data, and producing preliminary reports — tools intended to support, not supplant, expert decision-making.

Despite the promise, the authors emphasized the need for caution, transparency, and rigorous validation. “The ability to synthesize information from across the globe in a reliable manner would be very helpful,” Koopmans said. “But the question is — who checks the quality?”

As AI-generated outputs become increasingly difficult to distinguish from human analysis, assessing reliability and accuracy will be critical for public health officials and clinicians guiding outbreak response. Koopmans urged skepticism toward “black box” solutions and stressed that AI tools should be evaluated with the same rigor applied to new diagnostics.

Koopmans reported having no relevant financial conflicts of interest.