Artificial intelligence showed its power and screened out new antibiotics that could kill superbugs in an hour and a half.

Thanks to CTOnews.com netizen OC_Formula for clue delivery! CTOnews.com May 28, according to the results of a study published in the journal Nature Chemical Biology, researchers in Canada and the United States have used AI technology to discover a powerful antibiotic called abaucin. The antibiotic needs further testing before it can be used as a clinical treatment. Researchers say AI technology can greatly improve intelligence and efficiency and solve difficult problems in new drug development.

Antibiotics are drugs used to kill bacteria, but the lack of new antibiotic development in recent decades has led to bacterial diseases becoming increasingly difficult to treat as they evolve resistance to existing drugs. It is estimated that more than one million people die each year from infections resistant to antibiotic treatment.

The researchers targeted one of the toughest bacterial species, Acinetobacter baumannii, which can infect wounds and cause pneumonia and is one of three "serious" threats listed by the World Health Organization. CTOnews.com understands that the bacteria are resistant to a variety of antibiotics and can survive on surfaces and medical devices.

Dr Jonathan Stokes, from McMaster University, calls the bacterium "public enemy number one" because it is "very common" and is often found to be "resistant to almost all antibiotics."

To find new antibiotics, researchers first have to train AI. They took thousands of drugs with known chemical structures and manually tested them against Acinetobacter baumannii to see which slowed or killed it. This information is fed into the AI, which learns which chemical signatures can attack the troublesome bacteria.

They then asked AI to sift through a list of 6680 compounds with unknown effects. The AI generated a shortlist in an hour and a half. The researchers tested 240 compounds on the list in the lab and found nine potential antibiotics, one of which was the powerful antibiotic abaucin.

Laboratory experiments showed that abaucin was able to treat infected wounds in mice and was able to kill Acinetobacter baumannii samples from patients. The researchers say the next step is to refine the drug in the lab and then conduct clinical trials, with the first AI antibiotics expected to be prescribed by 2030.

Strangely, this experimental antibiotic had no effect on other bacteria, only Acinetobacter baumannii. Many antibiotics kill bacteria indiscriminately, and the researchers believe that abaucin's precision could make it harder for bacteria to develop resistance and potentially have fewer side effects.

In theory, AI can screen tens of millions of potential compounds, which is impractical for humans. "AI increases the speed at which we discover these new antibiotics that we desperately need and, ideally, reduces costs. Dr. Stokes said.

Professor James Collins, from MIT, said: "This finding further supports that we can use AI to accelerate and expand our search for novel antibiotics. He added: "I'm pleased to see that this work shows that we can use AI to help fight hard-to-handle pathogens such as Acinetobacter baumannii." "

References:

https://www.nature.com/articles/s41589-023-01349-8