Microbiology (Reading). 2025 Nov;171(11). doi: 10.1099/mic.0.001629.
ABSTRACT
Artificial intelligence (AI) and machine learning (ML) are reshaping microbiology, enabling rapid antibiotic discovery, resistance prediction and clinical diagnostics. For microbiologists, the goal is not to build new algorithms but to recognize when ML is appropriate, how to prepare data and how to interpret outputs responsibly. This primer takes that practical stance – driving the ML car rather than rebuilding the engine. At a high level, ML learns from complex patterns, often noisy data. In antibiotic discovery, ML models help identify compounds in biological data and design new ones from scratch using generative AI. In microbiome studies, where measurements are compositional, sparse and often confounded, ML helps uncover community structure and link taxa or functions to phenotypes. In pathogen genomics, supervised models map sequence-derived features (e.g. k‑mers, SNPs and gene presence/absence) to outcomes such as species identity, antimicrobial susceptibility or MIC. Unsupervised learning supports exploration, including clustering, latent gradients and dimensionality reduction for visualization. Across these settings, success hinges less on exotic architectures than on sound problem framing, careful preprocessing and experimental validation.
PMID:41231537 | DOI:10.1099/mic.0.001629
Share Evidence Blueprint
Save to Google Notes
Search Google Scholar
Save as PDF
⭐ My Revision List
