Overview

• Published June 4 in the Journal of the Royal Society Interface, the research demonstrates an AI-human closed-loop system applied to breast cancer drug discovery.
• Researchers selected GPT-4 hypotheses for low-toxicity, approved non-cancer drugs to streamline regulatory approval and accelerate clinical readiness.
• Laboratory tests of twelve GPT-4-suggested combinations identified three pairs that outperformed standard breast cancer treatments.
• A second cycle of AI refinement and testing highlighted simvastatin paired with disulfiram as a highly effective inhibitor of breast cancer cells.
• The most promising repurposed drug combinations are currently undergoing preclinical and clinical evaluation to determine their therapeutic potential.