Predicting antimicrobial resistance in Staphylococcus aureus using machine learning: Insights from a five-year...

Area of Science:

• Medical Microbiology
• Infectious Diseases
• Computational Biology

Background:

• Staphylococcus aureus is a major cause of community and hospital-acquired infections globally.
• Increasing antimicrobial resistance (AMR) poses a significant challenge to effective clinical management worldwide.
• Understanding local resistance patterns and leveraging predictive tools are crucial for combating S. aureus infections.

Purpose of the Study:

• To investigate the epidemiology and antimicrobial resistance trends of S. aureus in Saudi Arabia.
• To analyze multidrug resistance (MDR) patterns and identify factors contributing to resistance.
• To evaluate the utility of machine learning (ML) models in predicting antibiotic susceptibility for S. aureus.

Main Methods:

• Analysis of 18,003 microbiology reports from 2019-2024, identifying 2506 S. aureus isolates.
• Susceptibility testing against 31 antibiotics across 11 pharmacological classes.
• Development and evaluation of machine learning models (Random Forest, Logistic Regression, Gradient Boosting) for predicting antibiotic resistance.

Main Results:

• Wound and blood were the most common sources of S. aureus isolates.
• High resistance rates (>70%) were observed for beta-lactams, fluoroquinolones, and macrolides/lincosamides.
• Multidrug resistance (MDR) was present in 30% of isolates, while last-line antibiotics like vancomycin and linezolid showed preserved efficacy (<10% resistance).
• The Random Forest model demonstrated superior performance in predicting antibiotic susceptibility across most agents.

Conclusions:

• S. aureus remains a significant clinical threat in Saudi Arabia, characterized by high MDR rates.
• Last-line antibiotics retain effectiveness, highlighting their importance in treatment strategies.
• Machine learning offers a valuable tool for enhancing antimicrobial stewardship and informing clinical decision-making by predicting resistance patterns.