Recent advances in AI for antimicrobial resistance (AMR) prevention and control

Recent reviews highlight that AI is now being applied end‑to‑end across the AMR response:
enhancing diagnostic accuracy, guiding antimicrobial stewardship, improving surveillance, and accelerating drug and vaccine discovery.3 4 8

AI‑driven clinical decision support systems are being developed to optimise antibiotic prescriptions, predict bacterial resistance, and tailor treatment pathways, reducing misuse and overuse of antimicrobials.3 4 6 8

Machine‑learning models using whole‑genome sequencing data can predict resistance profiles and track the spread of resistant organisms, supporting real‑time AMR surveillance and early detection of emerging threats.3 4 6 8

In diagnostics, AI is being integrated with molecular and rapid testing platforms to enable faster pathogen identification and resistance prediction at the point of care, particularly important for low‑resource settings.3 4 6 8

Generative AI is now designing structurally novel antibiotic candidates at scale:
MIT and Broad Institute researchers used AI to screen tens of millions of molecules and identified new compounds active against multidrug‑resistant Neisseria gonorrhoeae and MRSA, with membrane‑disrupting mechanisms distinct from existing drugs.2 7

AI‑accelerated high‑throughput screening and small‑molecule prediction are compressing early antibiotic discovery timelines, allowing hundreds of thousands of preclinical candidates to be identified in hours instead of years.3 4 5

Beyond small molecules, AI and modern microbiology are enabling personalised bacteriophage therapies; algorithms help select and design phage ‘cocktails’ that target specific bacterial strains while reducing the risk of new resistance.1

AI is also being explored to support vaccine design against resistant bacteria, including using machine‑learning and RNA/protein structure models to identify antigens and engineer next‑generation vaccines that can indirectly curb AMR by preventing hard‑to‑treat infections.3

Policy‑oriented work (e.g., the BARDI framework and the Google DeepMind–Fleming Initiative collaboration) is defining strategic pillars—priority setting, data, evaluation, capacity, equity—for responsible, globally coordinated use of AI against AMR.3

Across healthcare systems, AI‑driven antimicrobial learning systems are being proposed to continuously learn from large‑scale clinical and microbiology data, adapt prescribing guidelines, and support sustainable, system‑level AMR control.3 4 6 8