This is a first: an artificial intelligence participated in the development of a new antibiotic described as overpowered.
Artificial intelligence will be invading in a growing number of sectors in the coming years and medicine could greatly benefit from this new technology.
Two researchers from MIT (Massachusettes Institute of Technology) have thus developed a machine learning algorithm which has recently managed to identify a new antibiotic compound that is more effective than the majority of what is currently known.
The algorithm in question has trained with digital libraries of pharmaceutical compounds and has carried out simulations with more than 2,500 molecules present in 1,700 drugs approved by the FDA and 800 natural products. The idea was to develop an antibiotic capable of overcoming bacterial strains resistant to all known antibiotics …
The antibiotic proposed by the algorithm was then successfully tested on two of the three test pathogens considered critical by the WHO. It was thus possible to eradicate Mucobacterium Tuberculosis, the virus responsible for tuberculosis, as well as Acinetobacter Baumannii which causes very resistant infections in mice.
With the help of such artificial intelligence, doctors and research could acquire a tool allowing them to have a head start on diseases by anticipating their adaptation to available drugs and by better controlling training. resistances.