In a landmark discovery, a machine learning based AI has been used to uncover powerful new antibiotics that can kill resistant bacterial strains.
Around 46,000 people dying each year in the UK alone from sepsis, with many cases being caused by antibiotic resistant bacteria that don't respond to treatment. The World Health Organisation has has identified several high-priority target pathogens that new antibiotics should target, but development of just one new drug can take years of research and millions of pounds in funding.
Researchers at the Massachusetts Institute of Technology (MIT) in the US have now successfully used an AI to discover new antibiotic drugs for the first time. The team trained the AI on a data set of known antimicrobial molecules and then set it loose on a vast pharmaceutical database to assess the potential of each drug as an antibiotic. The process dramatically cuts the cost and time needed for drug discovery.
Most antibiotics in use today work using the same basic methods, so researchers took the work a step further and tasked the AI with finding drugs that ranked highly in antibiotic potential but looked very different to existing known antimicrobial compounds. It produced several promising candidates, which were then tested on cultures of dangerous antibiotic-resistant bacteria.
The most potent of the new drugs, named Halcin, was able to kill a wide range of dangerous and resistant bacteria, including two of the three critical targets named by the World Health Organisation. Trials in mice showed that it was effective at clearing infections of antibiotic resistant straints.
The next step for the research is to search a much larger database of molecules using the AI in the hopes of finding even more antibiotic-resistant drugs. There are also plans to use the same technique to find drugs that can selectively kill specific infections while leaving the body's normal healthy bacteria alive.
Source: The Guardian
