Researchers at Cambridge University have used artificial intelligence to identify conserved viral targets and develop a vaccine candidate intended to protect across multiple coronaviruses. The candidate, called pEVAC-PS, underwent a Phase I clinical trial in which 39 healthy volunteers received the vaccine; according to published reports, the vaccine was well tolerated at all dose levels and induced immune responses that could be effective against several coronaviruses.
Why this matters
Conventional vaccines often target viral components that can change rapidly, so fast-mutating viruses like SARS-CoV-2 can partly evade prior immunity. The Cambridge team sought to overcome this limitation by focusing on antigens that rarely mutate, aiming for a broader and potentially more durable protection.
How the AI was used
The researchers trained models on genetic data collected from known sarbecoviruses — including sequences related to SARS-CoV-2 — to identify so-called "superantigens" that are both conserved and immunologically relevant. Those identified antigenic regions were used to design the pEVAC-PS vaccine candidate.
Results of the clinical study
pEVAC-PS was tested in 39 healthy volunteers in a Phase I trial. The report indicates that all dose levels were tolerable, no serious adverse events were observed, and the vaccine elicited immune responses that showed cross-reactivity against multiple coronaviruses. Details of the immunological findings are reported in a publication in the Journal of Infection.
Potential applications and limitations
The authors state that their platform could be adapted to produce broadly protective vaccines against other highly variable pathogens such as influenza and Ebola. However, they also stress that Phase I trials primarily assess safety and preliminary immunogenicity: while the results are encouraging, they do not guarantee success in later-stage trials.
Conclusion
The AI-assisted approach demonstrates a promising direction for vaccine design. The initial human data for pEVAC-PS suggest the method can produce candidates with cross-coronavirus activity and acceptable safety in small early trials, but further studies are required to confirm efficacy and durability of protection.
