A seasonal flu vaccine developed with the help of artificial intelligence by scientists at Flinders University is about to be tested in clinical trials across the United States.
The technology for this improved flu shot, developed by a team led by Flinders professor and research director of Vaxine, Nikolai Petrovsky, is being touted as the first human drug to be completely designed by AI.
The Flinders-based team created the AI program, called Search Algorithm for Ligands (SAM), and the technology behind the vaccine itself employs adjuvants, substances that enhance the body's immune response to a vaccine.
Petrovsky told Healthcare IT News the team had to teach the AI program a set of compounds known to activate the human immune system, and a set of compounds that don't work – the job of the AI was then to work out for itself what distinguishes a drug that worked from one that doesn't.
He explained they then developed another program, called the synthetic chemist, which generated trillions of different chemical compounds that they then fed to SAM so it could sift through all of them to find candidates it thought might be good human immune drugs.
"We then took the top candidates provided to us by SAM and synthesized them for the first time in the lab and tested them on human blood cells to see if they would work," he said. "This confirmed that SAM not only had the ability to identify good drugs, but in fact had come up with better human immune drugs than currently exist."
The team then took these drugs created by SAM into development with animal testing to confirm their ability to boost influenza vaccine effectiveness.
"This is now the candidate that is being tested in clinical trials in the U.S., just years after it was created by SAM, shortening the normal drug discovery and development process by potentially decades and hundreds of millions of dollars," Petrovsky noted.
The U.S. clinical trial, sponsored by the National Institute of Allergy and Infectious Diseases, part of the U.S. National Institutes of Health, will take about a year to complete and aims to recruit 240 healthy volunteers.
Petrovsky noted the biggest stumbling block was possibly the skepticism of his scientific colleagues and some local granting bodies that this approach would actually work.
"Notably in the original grant proposal we had five years to get the system working and deliver some initial candidates," he pointed out. "Instead, the AI was so powerful that we already had our final drug candidate within the first year of the grant so we ended up being more than four years ahead of ourselves."
He predicted AI would ultimately be the vehicle through which the vast majority of drugs of the future will be created.
"This is just a matter of time. We have shown that using such an approach we can screen 10 to the power 18 potential compounds – 10 with 18 zeros after it – in a matter of weeks," he said. "This is incomprehensible to even a large pharma company that may be able to screen several hundred thousands of compounds a year with a staff of hundreds."