AI invents more effective flu vaccine in world first, Adelaide researchers say
A “turbocharged” flu vaccine created by a computer with artificial intelligence in South Australia is set to be trialled in the United States.
Flinders University Professor Nikolai Petrovsky has told the ABC the computer running a program called Sam invented the new drug on its own, in what he claimed was a world first.
“It [Sam] has theoretical ability to acquire knowledge and then make new ideas,” Professor Petrovsky said.
“Obviously you have to train it or teach it.
“We took existing drugs that we know work, we took examples of drugs that don’t work or have failed.
“We essentially showed all of that to the AI program called Sam and then Sam came up with its own suggestion of what might be an effective adjuvant, which we then took and tested, and sure enough, it worked.”
Professor Petrovsky said a 12-month clinical trial of the vaccine was about to get underway in the US.
“Essentially, we have developed a technology that when added to existing flu vaccines makes them more effective,” he said.
Each year’s regular flu vaccine is an inoculation against four strains of the virus.
The chosen strains are decided by the World Health Organisation, according to which ones were prevalent in the previous northern or southern hemisphere flu seasons.
This year’s Australian vaccine for people aged over 65 contains a component in it which boosts their immune system.
Early and bad start to flu season
It comes amid an unusually early start to the flu season in Australia, which has seen 116,000 cases recorded and claimed more than 220 lives this year.
Australian chief medical officer Professor Brendan Murphy said this year’s season began in summer.
“We had a very low flu season last year, then a high what we call ‘inter-seasonal flu’ during summer, which has then morphed into an early flu season, which is really the earliest we’ve ever seen,” Professor Murphy said.
“What we don’t know is whether this season will fade out early and not be a very big season, or whether it’ll continue at its current level with another strain becoming predominant.
“It’s very hard to predict.”
Professor Murphy said he was unaware of the Flinders University program, but he welcomed the development.
“The complexity of drug and receptor interactions is so huge that’s it’s very definitely enhanced by IT systems and AI systems,” he said.
“So I think in general principles, it’s a very promising area to look at.”