MIT's Collins Lab has developed an AI-based method to identify new antibiotics, targeting drug-resistant superbugs like MRSA.

Researchers trained 20 deep-learning models on over 39,000 molecules, successfully pinpointing 3,600 potential non-toxic antibiotic compounds.

The AI utilized a Monte Carlo tree search to explain the antibiotic properties of molecules, enhancing understanding of their chemical substructures.

Following their discovery of halicin as a potential antibiotic, the team has made their AI model and data publicly available on GitHub to encourage collaboration.

The innovation addresses the critical public health issue of antibiotic resistance, which caused over a million deaths in 2019 according to WHO.

This breakthrough showcases the significant role AI could play in accelerating the discovery of new treatments for resistant bacterial infections.