Repurposing MALDI-TOF MS for effective antibiotic resistance screening in Staphylococcus epidermidis using machine...

Area of Science:

• Microbiology
• Computational Biology
• Clinical Diagnostics

Background:

• Staphylococcus epidermidis is a major cause of hospital-acquired infections.
• Current antimicrobial resistance testing methods are slow and expensive.
• Need for rapid and cost-effective resistance profiling.

Purpose of the Study:

• Develop predictive models for antibiotic resistance in S. epidermidis.
• Combine machine learning (ML) with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS).
• Validate the models for clinical application.

Main Methods:

• Utilized a large dataset of S. epidermidis isolates.
• Applied optimized ML models with feature selection.
• Employed Shapley Additive exPlanations (SHAP) for feature analysis.
• Performed external validation of the predictive models.

Main Results:

• Achieved high Area Under the Receiver Operating Characteristic Curve (AUROC) scores (0.80-0.95).
• Maintained high Area Under the Precision-Recall Curve (AUPRC) scores (up to 0.97).
• Identified significant protein biomarkers contributing to predictive power.
• Demonstrated robust model performance through external validation.

Conclusions:

• Presents a significant advancement in rapid antimicrobial resistance profiling.
• Offers a cost-effective solution for diagnosing nosocomial infections.
• Potential for application to other microbial pathogens in the future.