Researchers may have solved the world's drug-resistant bacteria problem using artificial intelligence

Peter Weber

Researchers at MIT reported Thursday that they have harnessed artificial intelligence to identify a completely new antibiotic compound that killed all but one of the antibacterial-resistant pathogens they tested it on. Drug-resistant bacteria are a large and growing problem, causing 2.8 million infections and 35,000 deaths in the U.S. each year and more in developing countries, STAT News reports. The computer learning model developed at MIT, described in the journal Cell, has the potential to identify many new types of antibiotics.

The researchers named the compound halicin, after HAL, the initially useful, eventually murderous sentient computer in 2001: A Space Odyssey. They also discovered eight more promising antibacterial compounds, two of which appear very powerful. They tried out halicin on mice and plan to work with a nonprofit or drugmaker to see if it's effective and safe in humans.

The MIT team's machine-learning model independently looked for certain properties — in this case, the ability to kill E. coli and not harm humans — among about 2,500 molecules in a drug repurposing database. Halicin was originally considered as a treatment for diabetes.

The model helps researchers find "leads among chemical structures that in the past we wouldn't have even hallucinated that those could be an antibiotic," Nigam Shah, a Standford professor of biomedical informatics who wasn't involved in the study, told STAT News. "To use a crude analogy, it's like you show an AI all the different means of transportation, but you've not shown it an electric scooter," he added. "And then it independently looks at an electronic scooter and says, 'Yeah, this could be useful for transportation.'"

In this case, said MIT medical engineering professor James Collins, the platform "revealed this amazing molecule which is arguably one of the more powerful antibiotics that has been discovered." Collins is a senior author of the study along with MIT computer scientist Regina Barzilay. You can read more about their work at STAT News and MIT.

More stories from theweek.com
CNN analyst: Republicans 'may regret' hoping Sanders wins nomination
The stunning Southern Baptist controversy over Donald Trump and Russell Moore, explained
White House officials are reportedly hoping to scale back surveillance powers