Staphylococcus aureus resistance to non-β-lactam antibiotics: global genomic epidemiology and machine learning feasibility assessment - PubMed
a day ago
- #Antimicrobial Resistance
- #Machine Learning
- #Genomic Epidemiology
- Staphylococcus aureus resistance to non-β-lactam antibiotics is increasing, necessitating genomic and phenotypic studies.
- 112,360 S. aureus genomes were analyzed, with 111,350 meeting quality standards, sourced from 137 countries.
- Resistance genes (e.g., fosB, bcrAB) were identified, with fosB prevalent in 65.3% of genomes, especially in clinical isolates.
- Glycopeptide MICs (vancomycin, teicoplanin) were low, while daptomycin showed variability; fosfomycin and bacitracin had limited phenotypic data.
- FosB resistance increased by 0.20% annually, with geographic trends showing high prevalence in North America, Europe, and South America.
- Machine learning models had moderate success predicting daptomycin MICs (R² = 0.49) but failed for glycopeptides due to low MIC variability.
- The study underscores the need for integrated genomic and phenotypic surveillance to improve resistance prediction and management.