Label-Free Differentiation of Antimicrobial Resistance Groups Using Raman Spectroscopy.

Area of Science:

• Spectroscopy and Machine Learning Applications in Clinical Microbiology
• Antimicrobial Resistance (AMR) Detection and Diagnostics

Background:

• Antimicrobial resistance (AMR) poses a significant global health challenge.
• Accurate and rapid identification of resistance mechanisms is crucial for effective treatment.
• Current methods for AMR detection can be time-consuming, delaying appropriate patient care.

Purpose of the Study:

• To evaluate Raman spectroscopy's performance in discriminating between bacterial resistance classes.
• To assess different Raman spectroscopic methods and machine learning algorithms for AMR detection.
• To develop a decision-tree approach for species classification and resistance differentiation.

Main Methods:

• Investigated two Raman spectroscopic methods: UVRR and 785 nm excitation.
• Employed four machine learning algorithms: PCA-LDA, PLS-DA, PCA-SVM, and PCA-RF.
• Tested on common clinical isolates including *Escherichia coli*, *Klebsiella pneumoniae*, and *Enterococcus faecium*.

Main Results:

• Raman spectroscopy achieved high accuracy in species classification.
• 785 nm excitation combined with PCA-SVM and PLS-DA demonstrated superior performance in differentiating susceptible from resistant strains.
• These methods accurately identified ESBL and CRE isolates, showing stability and suitability for hospital settings.

Conclusions:

• Raman spectroscopy, particularly with 785 nm excitation and specific machine learning models, is a promising tool for rapid AMR diagnostics.
• This approach can aid in microbiological documentation and provide critical AMR-related information for infectious disease management.
• Further development of Raman spectroscopy holds potential for revolutionizing clinical microbiology diagnostics.