# Integrating MALDI-TOF Mass Spectrometry and Machine Learning for Rapid and Clinically Relevant Differentiation of MRSA and MSSA

## TL;DR

This study shows that combining MALDI-TOF mass spectrometry with machine learning can quickly and accurately distinguish MRSA from MSSA, aiding faster antimicrobial decisions.

## Contribution

The novel use of machine learning with MALDI-TOF MS data to differentiate MRSA and MSSA in clinical settings.

## Key findings

- A Random Forest classifier achieved 81.3% accuracy in differentiating MRSA and MSSA using MALDI-TOF MS data.
- The model showed high precision for MRSA (95.5%) and excellent recall for MSSA (98.1%).
- Principal component analysis revealed partial but consistent separation between MRSA and MSSA isolates.

## Abstract

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) is routinely used in clinical microbiology for rapid species identification; however, its potential to support early antimicrobial decision-making remains under active investigation. Rapid discrimination of methicillin-resistant Staphylococcus aureus (MRSA) and methicillin-susceptible Staphylococcus aureus (MSSA) at the time of identification could facilitate earlier optimization of antimicrobial therapy and infection control measures. In this study, MALDI-TOF MS spectral data were analyzed to evaluate supervised machine learning–based differentiation of MRSA and MSSA. A total of 91 S. aureus isolates (37 MRSA and 54 MSSA) were included, with methicillin susceptibility determined by the cefoxitin disk diffusion test according to EUCAST guidelines and used as the reference standard. MALDI-TOF MS spectra were acquired following standard on-plate extraction, subjected to quality control, and preprocessed prior to analysis. Principal component analysis demonstrated partial but consistent separation between MRSA and MSSA isolates. A Random Forest classifier was trained and validated using stratified 10-fold cross-validation, achieving an overall classification accuracy of 81.3% and a receiver operating characteristic area under the curve of 0.916. Class-specific analysis revealed high precision for MRSA (95.5%) and excellent recall for MSSA (98.1%). These findings indicate that MALDI-TOF MS combined with machine learning can provide clinically relevant information for rapid MRSA/MSSA differentiation and may serve as a cost-free decision-support approach in routine clinical microbiology workflows, complementing standard phenotypic susceptibility testing.