AI-driven approaches to vaccine design are rapidly expanding, with the number of research papers on AI-powered vaccine development increasing by 5,200% over the past decade, according to data from Elsevier’s Scopus database. This coincides with the first human clinical trial of a vaccine whose key components were designed entirely by AI.
The AI-designed vaccine, developed by researchers at the University of Cambridge and its spinout DIOSynVax, uses “superantigens” that aim to provide lasting protection even if multiple Sarbeco coronaviruses mutate.
Researchers say moving forward to human trials has demonstrated the safety of a fundamentally new approach to vaccine design using computer simulation.
To design the antigen, the researchers used all available genetic sequence data for Sarbeco coronaviruses recorded by global virus surveillance programs. They used machine learning to analyze sequence data and design antigens that protect against this group of viruses, including variants that have not yet emerged.
sarbecocoronavirus
Sarbecocoronaviruses are a large group of viruses that occur in nature, primarily in bats, and are known for their zoonotic ability to spread to humans. This subgenus is best known for containing both SARS-CoV-1 and SARS-CoV-2, the viruses that cause the COVID-19 pandemic.
AI-designed vaccines represent a major change from traditional vaccines, which use antigens from specific virus strains or variants that have already been detected in humans. This typical approach to vaccine design requires updating the vaccine every year as the virus mutates.
“We have shifted vaccine development from being reactive to being proactive. Our vaccine continues to provide protection even if the virus mutates into new strains,” Professor Jonathan Heaney, the study’s scientific lead, said in a press release.
The vaccine’s Phase 1 trial involved 39 healthy volunteers between the ages of 18 and 50 who had received two or three previous COVID-19 vaccinations and had no recent known infection. Participants received escalating doses of the vaccine on days 0 and 28. The results were recently infection journal.
The DNA vaccine was administered by needle-free intradermal injection. This method of administration provides an alternative for people who have a fear of needle injections. It could also make it easier to administer vaccinations in settings where traditional injections are difficult.
The vaccine was well tolerated and did not raise any safety concerns. The majority of adverse events reported were mild or moderate. Fewer events were reported after subsequent doses, suggesting good tolerability for repeated injections.
Although the safety profile was promising, the effects of the vaccine on the immune system were modest and variable. Neutralizing antibodies against the delta and omicron variants were slightly increased, but activity against the ancestral Wuhan strain and SARS-CoV-1 was limited.
Differences between participants in previous SARS-CoV-2 exposure and vaccination history may have influenced these results. Recruitment for this study occurred during a flurry of Omicron infections and booster campaigns, complicating interpretation of the vaccine’s effectiveness.
The researchers confirmed with peptide microarray analysis that the vaccine targets immunity to conserved regions of the virus. It revealed antibody binding to receptor binding domain epitopes, including a wide range of known reactive antibody sites. Although such constraints are not necessarily strong, in vitro neutralization, these same binding events are associated with significant effects in vivo Protection activities through Fc-mediated mechanisms.
Breaking the reactive vaccine development cycle
This new class of AI-designed universal vaccines has the potential to take vaccine development out of the crisis cycle and protect against many variants at once.
Traditional vaccine research tends to follow trends. Funding and development increases when threats become visible and decreases when immediate risks decrease.
Recent data from Scopus, Elsevier’s peer-reviewed academic literature summary and citation database, shows how Ebola vaccine research is following this trend (Figure 1).
In 2015, the number of research papers on Ebola vaccine development peaked, coinciding with the 2014-2016 Ebola crisis in West Africa. In 2020-2021, around the time the first Ebola vaccine was developed, publications increased again. zaire ebola virus Approved. Since then, publications have declined, perhaps reflecting less urgency, even though other Ebola virus species continue to pose a threat.
Figure 1: Trends in the number of published research papers on Ebola vaccine development in the Scopus database. Credit: Data provided by Elsevier.
Recently, growing concern about the possibility of H5N1 transmission from birds to mammals and humans has led to a surge in H5N1 vaccine research (Figure 2). After fluctuations over the past decade, the number of H5N1 vaccine publications increased sharply in 2024 and 2025. This is believed to be part of the response to the dairy cow outbreak reported in March 2024 in several U.S. states.
Figure 2: Trends in the number of research papers published on H5N1 vaccine development in the Scopus database. Credit: Data provided by Elsevier.
While traditional vaccine designs often respond to visible threats, new vaccine technologies are designed to proactively address outbreaks and unknown ramifications. Universal vaccine technology could end the “reactive” vaccine development system that keeps up with virus evolution.
“This new class of universal vaccines is future-proofed. It has the potential to protect against many variants at once, as well as from related viruses that have not yet emerged and have not spread to humans,” said Professor Saul Faust, lead researcher on the trial. “If we can develop and clinically advance this new type of vaccine before the virus outbreak begins, it could save millions of lives, avoid lockdowns and keep the economy afloat.”
Power of AI vaccine research
The completion of the first human trial of an AI-designed vaccine is the culmination of growing interest in the use of computational biology and machine learning in vaccine development.
According to data from Scopus, only four papers were published on AI-powered vaccine development in 2015. Ten years later, this number had jumped to 212 papers. India, the United States, and China are leading the growth in AI vaccine research, accounting for more than half of the papers published on the subject in the Scopus database since 2015 (Figure 3).
Figure 3: Number of research articles on AI and vaccine development in the Scopus database by country (top 15 countries shown). Credit: Data provided by Elsevier.
The proliferation of AI-designed vaccine publications signals a major shift toward more predictive immunology and technologies that can provide broad and lasting viral protection.
Further development is required before a sarbecocoronavirus vaccine can be rolled out to the public. Larger phase 2 trials in broader and more diverse populations are needed to confirm that vaccines generate strong and broadly protective immunity.
“Given the continued risk of SARS-CoV-2 evolution and the potential for future zoonotic coronaviruses, the development of broadly protective vaccines remains a key global health priority,” the researchers concluded.
reference: Munro AP, Ferrari M, Kinsley R, et al. Phase I needle-free dose-escalation clinical trial of pEVAC-PS, a pan-sarbecovirus vaccine candidate. J Infect. 2026;92(6). Doi: 10.1016/j.jinf.2026.106759
