AI Repurposes Existing Drugs to Target Antibiotic-Resistant Infections
Machine-learning screening is pointing researchers toward already-approved compounds that can be redirected against drug-resistant bacteria.
Researchers are turning to machine learning to find new uses for drugs that are already on the market, aiming them at bacteria that have stopped responding to standard antibiotics. The approach, reported by Inside Precision Medicine, treats drug repurposing as a faster route than designing new molecules from scratch. With resistant infections rising and the antibiotic pipeline thin, the strategy has drawn fresh attention.
The core idea is to let algorithms sift through libraries of approved compounds and predict which ones might interfere with bacterial survival, then confirm those hits in the lab. Because these drugs have already cleared safety testing in humans, any promising candidate can move toward clinical trials far more quickly than a novel antibiotic, which typically takes a decade or more to develop. That speed matters when resistance can render a frontline treatment useless within years.
Antibiotic resistance is not a distant threat. The World Health Organization has flagged it as one of the top global public health concerns, and drug-resistant infections are linked to millions of deaths annually worldwide. Yet major pharmaceutical companies have largely abandoned antibiotic development because the economics are poor: new drugs are used sparingly and priced low, so returns rarely justify the investment. Repurposing sidesteps part of that problem by starting from compounds that already exist.
AI's role here is not magic. Machine-learning models trained on chemical structures, gene expression data and known drug-bacteria interactions can rank thousands of candidates in silico, narrowing the list that scientists then test at the bench. Several academic groups and biotech startups have used similar methods to surface antivirals and cancer drugs with unexpected antibacterial activity. The technique works best as a filter, cutting experimental workload rather than replacing it.
The open question is whether repurposed drugs will hold up in patients. A compound that kills resistant bacteria in a petri dish may fail in the body because of dosing, toxicity or interactions with other medicines. Researchers will need to run trials that prove real clinical benefit, not just laboratory activity. For now, the field is watching which of these AI-flagged candidates advance to human testing, and whether regulators and funders treat repurposing as a serious answer to a slow-moving crisis.