AI reveals how H5N1 virus is evolving to evade human immunity

The H5N1 avian influenza virus has infected birds and mammals around the world. As of June 2025, 70 people have been infected, and 1 person has died in the United States. A new analysis suggests that the virus is evolving clever strategies. Using artificial intelligence tools, researchers at the University of North Carolina at Charlotte (UNCC) analyzed thousands of viral proteins and found that their bonds to protective antibodies have weakened over time. 

Newer versions of the virus have improved their ability to evade the natural defenses of the human immune system. "The virus has certainly mutated away from what we saw a decade ago," said UNCC computational biologist Colby T. Ford, Ph.D., who led the study. "They don't even look the same." 

These adaptations increase the pandemic potential of the virus, he said, and candidate vaccines developed 10 years ago may not be efficacious against the contemporary strains of the virus. "This has the potential to be bad."

Together with Shirish Yasa, B.S., who worked on the study as an undergraduate and is now a student at Eastern Virginia Medical School, Ford presented the findings at ASM Microbe 2025, the annual meeting of American Society for Microbiology, in Los Angeles. 

The researchers first collected data on more than 1,800 H5N1 proteins. They used AlphaFold 3, an artificial intelligence protein folding system, to predict the complicated structures of the viral proteins. Then, using physics-based modeling systems, they tested how well 11 immune antibodies-collected from both people and mice-attached to the proteins. 

Better bonds mean better protection, said Ford, but the analysis revealed that over the years the binding has been weakening. "Antibody performance is waning as we get to the newer isolates that we're seeing."

The group has also been using large datasets focused on H5N1 to better connect clades of the virus to specific transmission channels.

We can see that there are distinct clades with very different paths in terms of transmission between hosts."

Colby T. Ford, Ph.D., UNCC computational biologist 

The group recently connected the H5N1 death of a person in Louisiana to a clade that can pass directly from bird to human, without having to go through another animal. 

These analyses show how the virus is finding strategies to evade the immune system, Ford said, but they also show how AI and computational modeling can help researchers track the evolution of the virus and, potentially, design more effective antibodies. In a preprint, the group has described an approach that uses molecular information from new and emerging strains to design effective, targeted treatments. 

"Can we start to generate novel therapeutics based off those strains? The answer is yes, and we can do it fairly quickly with the AI pipeline we've built," Ford said.