Artificial intelligence is already helping scientists discover new drugs, predict protein structures, and accelerate medical research. Now it has reached another milestone: helping design a vaccine that has been tested in humans for the first time.

Researchers have published results in the Journal of Infection from a Phase I clinical trial of pEVAC-PS, an experimental vaccine designed using AI and computational modelling.  

Their goal is ambitious: create a single vaccine capable of protecting against not just today's coronaviruses, but future ones that have yet to emerge.

That might sound impossible, but it addresses one of the biggest challenges in vaccine development.

Traditional vaccines are built to recognise a specific virus. They work brilliantly, until the virus changes. We've seen this repeatedly with influenza and, more recently, with COVID-19, where new variants emerged faster than vaccines could be updated and distributed.

This team took a very different approach.

Instead of focusing on a single virus, they used AI and computational screening tools to analyse thousands of related coronaviruses from the group that includes SARS, COVID-19, and many bat viruses that have the potential to jump into humans in the future. The algorithms searched for regions of these viruses that remain remarkably consistent across strains and over time. These conserved regions are often critical to the virus's survival, making them less likely to mutate.

The result was a synthetic vaccine antigen designed to train the immune system to recognise shared features across an entire family of viruses rather than one specific strain.  

Unlike the mRNA vaccines many people became familiar with during the COVID-19 pandemic, pEVAC-PS is a DNA vaccine. DNA vaccines have several practical advantages. They are generally more stable, easier to store, and less dependent on ultra-cold refrigeration.  

The vaccine was also delivered without a needle. Instead, researchers used a high-pressure device that injects the vaccine through the skin using a fine stream of liquid. Needle-free delivery reduces sharps waste, can improve vaccine acceptance, and may simplify large-scale deployment during future outbreaks.

The Phase I trial enrolled 39 healthy volunteers who had previously received COVID-19 vaccines. The primary purpose was not to determine whether the vaccine prevented disease, but to answer two fundamental questions: Is it safe? And does it trigger an immune response?

The good news is that the vaccine performed well from a safety perspective. Across four different dose levels, researchers found no serious safety concerns and no evidence that higher doses created more significant side effects. Overall, the vaccine was described as safe and well tolerated.

Detailed analysis showed that some antibodies targeted areas of the virus that scientists believe are important for broad protection across the coronavirus family. This is an encouraging sign because it suggests the AI-guided design strategy is identifying biologically meaningful targets.

However, the overall immune response was modest. Antibody levels generally remained close to pre-existing levels, and increases in neutralising antibodies were relatively small. In other words, it did not yet generate the strong, broad protection researchers ultimately hope to achieve.

That doesn't mean the approach has failed.

Phase I trials are designed primarily to establish safety and provide an early signal of biological activity. In this case, the study demonstrated that a vaccine designed using computational methods and AI can be safely administered to humans and can stimulate immune responses against conserved coronavirus targets. That's an important proof of concept.

The bigger story may be what this means for the future of vaccine development.

For decades, vaccine design has largely been reactive: a new virus appears, scientists study it, and then create a vaccine. AI offers the possibility of becoming more proactive. By analysing huge datasets of viral genomes, identifying stable targets, and predicting how viruses might evolve, future vaccines could be designed to provide protection against threats before they emerge.

We're not there yet. A universal coronavirus vaccine remains several years away, and larger clinical trials will be needed to determine whether this approach provides meaningful protection in the real world. But this study offers an intriguing glimpse of what next-generation vaccine design could look like.

Rather than chasing the next pandemic, AI may eventually help us prepare for it before it arrives. 

LISTEN ABOVE

Artificial intelligence is already helping scientists discover new drugs, predict protein structures, and accelerate medical research. Now it has reached another milestone: helping design a vaccine that has been tested in humans for the first time.
Researchers have published results in the&nbsp;<a href="https://www.journalofinfection.com/article/S0163-4453(26)00084-8/fulltext">Journal of Infection</a> from a Phase I clinical trial of pEVAC-PS, an experimental vaccine designed using AI and computational modelling. &nbsp;
Their goal is ambitious: create a single vaccine capable of protecting against not just today's coronaviruses, but future ones that have yet to emerge.
That might sound impossible, but it addresses one of the biggest challenges in vaccine development.
Traditional vaccines are built to recognise a specific virus. They work brilliantly, until the virus changes. We've seen this repeatedly with influenza and, more recently, with COVID-19, where new variants emerged faster than vaccines could be updated and distributed.
This team took a very different approach.
Instead of focusing on a single virus, they used AI and computational screening tools to analyse thousands of related coronaviruses from the group that includes SARS, COVID-19, and many bat viruses that have the potential to jump into humans in the future. The algorithms searched for regions of these viruses that remain remarkably consistent across strains and over time. These conserved regions are often critical to the virus's survival, making them less likely to mutate.
The result was a synthetic vaccine antigen designed to train the immune system to recognise shared features across an entire family of viruses rather than one specific strain. &nbsp;
Unlike the mRNA vaccines many people became familiar with during the COVID-19 pandemic, pEVAC-PS is a DNA vaccine. DNA vaccines have several practical advantages. They are generally more stable, easier to store, and less dependent on ultra-cold refrigeration. &nbsp;
The vaccine was also delivered without a needle. Instead, researchers used a high-pressure device that injects the vaccine through the skin using a fine stream of liquid. Needle-free delivery reduces sharps waste, can improve vaccine acceptance, and may simplify large-scale deployment during future outbreaks.
The Phase I trial enrolled 39 healthy volunteers who had previously received COVID-19 vaccines. The primary purpose was not to determine whether the vaccine prevented disease, but to answer two fundamental questions: Is it safe? And does it trigger an immune response?
The good news is that the vaccine performed well from a safety perspective. Across four different dose levels, researchers found no serious safety concerns and no evidence that higher doses created more significant side effects. Overall, the vaccine was described as safe and well tolerated.
Detailed analysis showed that some antibodies targeted areas of the virus that scientists believe are important for broad protection across the coronavirus family. This is an encouraging sign because it suggests the AI-guided design strategy is identifying biologically meaningful targets.
However, the overall immune response was modest. Antibody levels generally remained close to pre-existing levels, and increases in neutralising antibodies were relatively small. In other words, it did not yet generate the strong, broad protection researchers ultimately hope to achieve.
That doesn't mean the approach has failed.
Phase I trials are designed primarily to establish safety and provide an early signal of biological activity. In this case, the study demonstrated that a vaccine designed using computational methods and AI can be safely administered to humans and can stimulate immune responses against conserved coronavirus targets. That's an important proof of concept.
The bigger story may be what this means for the future of vaccine development.
For decades, vaccine design has largely been reactive: a new virus appears, scientists study it, and then create a vaccine. AI offers the possibility of becoming more proactive. By analysing huge datasets of viral genomes, identifying stable targets, and predicting how viruses might evolve, future vaccines could be designed to provide protection against threats before they emerge.
We're not there yet. A universal coronavirus vaccine remains several years away, and larger clinical trials will be needed to determine whether this approach provides meaningful protection in the real world. But this study offers an intriguing glimpse of what next-generation vaccine design could look like.
Rather than chasing the next pandemic, AI may eventually help us prepare for it before it arrives.&nbsp;
LISTEN ABOVE

News, opinion, analysis, lifestyle and entertainment – we’ve got your Sunday morning listening covered with The Sunday Session with Francesca Rudkin on Newstalk ZB.

The Sunday Session with Francesca Rudkin

Dr Michelle Dickinson: nanotechnologist on the world's first vaccine designed by AI

https://api.omny.fm/orgs/7784f840-c291-422a-924b-ad9000bbad71/clips/9a69f4f0-5780-45e2-bd86-b4690004a6bc/transcript