Same-day antimicrobial susceptibility test using acoustic-enhanced flow cytometry visualized with supervised machine...

Area of Science:

• Clinical microbiology
• Bioinformatics
• Medical technology

Background:

• Traditional antimicrobial susceptibility testing (AST) is time-consuming, delaying critical treatment decisions.
• Rapid AST methods are needed to improve patient outcomes and combat antimicrobial resistance.

Purpose of the Study:

• To assess the feasibility of using machine learning (ML) to analyze data from a novel flow cytometer-assisted antimicrobial susceptibility test (FAST) method.
• To determine if ML-FAST can provide rapid and accurate antimicrobial susceptibility results compared to standard methods.

Main Methods:

• Developed a machine-learning algorithm to analyze FAST data, comparing results against broth microdilution (BMD) AST.
• Tested various bacterial strains including *Escherichia coli*, *Klebsiella pneumoniae*, and *Staphylococcus aureus*.
• Applied the ML-FAST method to isolates directly from blood cultures for same-day analysis.

Main Results:

• The ML algorithm successfully classified bacterial responses to antimicrobials and predicted inhibitory concentrations.
• High concordance was observed between ML-FAST and BMD results (91% categorical agreement, 100% essential agreement for reference strains).
• Same-day AST results were achieved for clinical isolates, demonstrating the method's potential for rapid diagnostics.

Conclusions:

• The integration of machine learning with the FAST method enables same-day antimicrobial susceptibility testing.
• This innovative approach holds significant potential for guiding early antimicrobial treatment, enhancing stewardship, and facilitating resistance detection.