AI Identifies New Antibiotic Candidate That Eliminates 99% of MRSA Superbugs

A reinforcement learning-based AI model called 'SynteMol-RL' identified a new antibiotic candidate that reduced bacterial counts by more than 99% in a mouse model of methicillin-resistant Staphylococcus aureus (MRSA) infection. The model predicted candidates from approximately 46 billion molecules, and 13 of the 79 compounds actually synthesized—about 16%—showed potent antibacterial activity in vitro. The most promising candidate, 'Syntecin,' reduced bacterial counts in MRSA-infected skin from approximately 6.4 billion per gram to 51 million per gram compared with the control group when administered topically, demonstrating its potential for early treatment.