Integrated Analysis for Identification, Phenotyping, and Antimicrobial Susceptibility Testing (AST) of Bacteria Using Mass Spectrometry, Machine Learning, and Multi-omics Analysis

Abstract

Antimicrobial resistance (AMR) is gradually becoming a global public health problem. Rapid and cost-effective identification of AMR bacteria is the key to guiding the therapeutic management of bacterial infections/diseases. Mass spectrometry (MS) has been progressively adopted in clinical laboratories, especially for species identification. A series of supervised machine learning models have been systematically studied and have been shown to have great potential in strain-level typing. In the meantime, metabolites and lipids have been proven to facilitate pathogen typing, especially for differentiating SNP variants. More strikingly, the integration of multi-omics data has moved MS-based bacterial typing beyond identification and antimicrobial susceptibility testing (AST) to understanding the molecular mechanisms of AMR evolution.