A new fusion strategy for rapid strain differentiation based on MALDI-TOF MS and Raman spectra

Area of Science:

• Microbiology
• Analytical Chemistry
• Bioinformatics

Background:

• Accurate bacterial subspecies typing is crucial for managing disease outbreaks and guiding effective treatment strategies.
• While physicochemical spectroscopy offers rapid analysis, current identification accuracy for bacterial strains remains insufficient.
• Existing methods struggle with the fine-level differentiation required for precise diagnostics.

Purpose of the Study:

• To develop a novel, high-accuracy, and rapid bacterial strain differentiation strategy.
• To enhance pathogenic bacterial subtyping by fusing data from MALDI-TOF MS and Raman spectroscopy.
• To improve diagnostic capabilities for critical bacterial pathogens.

Main Methods:

• A feature-extractor-based fusion strategy integrating matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) and Raman spectroscopy.
• Application of machine learning algorithms including k-nearest neighbors (KNNs), support vector machines (SVMs), and artificial neural networks (ANNs) for data analysis.
• Validation using a panel of key pathogens: *Staphylococcus aureus*, *Klebsiella pneumoniae*, *Escherichia coli*, and *Acinetobacter baumannii*.

Main Results:

• The fusion approach achieved 100% identification accuracy for the tested bacterial pathogens using KNNs, SVMs, and ANNs.
• Significantly improved the identification accuracy of *Acinetobacter baumannii* from 87.67% to 100% compared to MALDI-TOF MS alone.
• Demonstrated reliable and rapid bacterial analysis and identification within 24 hours.

Conclusions:

• The developed fusion-assisted machine learning strategy effectively combines MALDI-TOF MS and Raman spectroscopy for superior pathogenic bacterial subtyping.
• This integrated approach offers a powerful tool for rapid and accurate bacterial identification, crucial for clinical diagnostics and outbreak response.
• The study highlights the potential of multi-modal spectroscopic data fusion for advancing microbial identification technologies.