A paper published online in Cell Reports Medicine in November by the University of Cambridge and other institutions introduced an AI-assisted design of an avian influenza vaccine candidate.
The results from animal testing are good. In mice and ferrets, it showed broad-spectrum immunity and protection, superior to a preparation based on a WHO candidate vaccine virus used as a control in the study.
The key is the antigen. Traditional vaccines usually start from virus strains that exist in nature, but this time the broad-spectrum antigen is artificially designed and does not exist in nature. The role of AI here is to speed up antigen design and quickly find solutions that can cover multiple strains at the same time in the huge space of possibilities.
The difficulty in making an avian influenza vaccine is that the virus changes quickly, and the circulating strains may be different every year. Traditional vaccines are often produced in advance based on the predicted strains. If you guess wrong, the protection will be reduced. The idea of a broad-spectrum vaccine is to find targets common to multiple strains and cover more variants with one shot.
Ferrets are the animal model closest to humans in influenza research, so its effectiveness in ferrets is a relatively weighty signal.
But don't jump to conclusions. The paper team themselves also said that the protective effect of this vaccine on humans has not yet been confirmed. The next step is to advance human clinical trials, with the goal of preparing for future emergency use.