AI just discovered an antibiotic that can kill a dangerous superbug

Researchers have tapped into the power of Artificial Intelligence to help discover a new antibiotic that can defeat the spread of a deadly drug-resistant superbug. But how was this discovery made and exactly which dangerous species of bacteria does it fight?

Roughly 1.2 million people die each year as a result of drug-resistant bacterial infections according to the University of Oxford but there is one major superbug that scientists say could prove to be extremely problematic if we don’t find a way to stop it from spreading. 

The bacteria in question is Acinetobacter baumannii, and while you may have not heard of it before, this dangerous little pathogen was listed by the World Health Organization as one of the top twelve bacterial diseases we need to worry about way back in 2017.  

Acinetobacter baumannii is a dangerous pathogen because it is known to cause serious blood, lung, and urinary tract infections, and can also live in a patient without causing infection or symptoms according to the Centers for Disease Control and Prevention. 

Survivability is a big worry if you catch an infection of Acinetobacter baumannii and the American Society for Microbiology reported that between 26% and 55% of patients will die from complications related to the pathogen that cannot be killed. At least until now.

Scientists in the United States and Canada were able to use Artificial Intelligence (AI) to help combat the threat of Acinetobacter baumannii after an advanced model capable of learning chemical features was released on a list of 7750 possibly helpful molecules. 

“We had a whole bunch of data that was just telling us about which chemicals were able to kill a bunch of bacteria and which ones weren’t,” McMaster University’s Gary Liu told The Guardian Maya Yang about what information the research team had to work with. 

“My job was to train this model, and all that this model was going to be doing is telling us essentially if new molecules will have antibacterial properties or not,” Liu added. 

The results of the AI’s work were published in the journal Nature Chemical Biology and the results were nothing short of astounding. The AI model was able to return a list of possible molecules that could combat Acinetobacter baumannii within a few hours.

Researchers then tested 240 of the molecules with the highest prediction of being able to inhibit the growth of Acinetobacter baumannii and out of that group they found nine compounds that had a very strong possibility of being an effective treatment. 

All nine of the compounds were then assessed until the researchers found one that they believed could be used effectively to contain Acinetobacter baumannii. That compound was abaucin and it proved to be extremely good at killing Acinetobacter baumannii.

The researchers tested abaucin’s ability to treat an Acinetobacter baumannii infection and found that the pathogen could be controlled in a mouse wound model, leading the team to declare that the compound could be used as an effective antibiotic.

“This work validates the benefits of machine learning in the search for new antibiotics” McMaster University’s Dr. Jonathon Stokes explained to the Guardian. Dr. Stokes lead the team that discovered abaucin and said AI in medicine could be revolutionary.

“Using AI, we can rapidly explore vast regions of chemical space,” Dr. Stokes explained to The Guardian, “significantly increasing the chances of discovering fundamentally new antibacterial molecules.”

In a separate interview with BBC News, Dr. Stokes explained that the work had only really begun now that abaucin had been discovered and said that the next steps would be to perfect the new drug in the laboratory before starting clinical trials.

Interestingly, the BBC reported that abaucin had no effect on other bacterium and was only a useful tool in combating Acinetobacter baumannii, something that surely shows the potential for Artificial Intelligence to target solutions humans may never have found.