Enhancing surveillance and early warning of infections and antimicrobial resistance using machine learning and deep learning: A systematic review

This review shows that artificial intelligence (AI), machine learning (ML), and deep learning (DL) have significant potential to transform antimicrobial resistance (AMR) management by improving surveillance, accelerating drug discovery, and enhancing clinical decision-making. Across 42 studies, AI tools demonstrated high performance—achieving ~90% accuracy in detecting resistance genes, ~91% accuracy in outbreak prediction, and reducing detection times by ~20% compared to conventional methods. In drug discovery, generative AI approaches (e.g., leading to compounds like halicin) have rapidly identified thousands of candidate antimicrobials with high in vitro efficacy, while clinical decision support systems reduced inappropriate antibiotic use in ICUs by ~30%. However, key limitations remain, including biased datasets (skewed toward high-income regions), limited model transparency, high computational demands, and a gap between discovery and clinical implementation, with few AI-derived compounds reaching trials. Overall, while AI offers powerful tools for precision diagnostics and innovation in AMR, its real-world impact depends on addressing data equity, validation, and responsible implementation.