AI-Driven Antibiotic Discovery: Addressing Antimicrobial Resistance Through Machine Learning

Abstract

Antibiotic resistance is a growing global issue, owing to the fast evolution of infections and the lessened efficacy of existing therapies. Unlike conventional medication development, antibiotics have the unique problem of being left behind as resistance develops. This has led to renewed interest in artificial intelligence (AI) and machine learning (ML) as approaches to expedite antibiotic discovery, particularly in the context of a slow and costly development process. This work reviews the increasingly widespread application of AI toward identifying antimicrobial peptides and small molecule drugs. These include prediction of antimicrobial activity, representation of compounds, assessment of drug-likeness, modelling of resistance mechanisms and de novo design of molecular classes. We also explore how open scientific principles, including reproducibility, openness, and data sharing, can be incorporated to accelerate preclinical research. We end by discussing emerging trends and future directions in antibiotic discovery, emphasizing how advances in machine learning are revolutionizing the field to tackle this urgent global challenge.