Microbial and antimicrobial resistance diagnostics by gas sensors and machine learning

This perspective discusses the potential of analyzing the volatilome of biofluids (e.g., urine, blood, sputum, bronchial alveolar lavage or stool), which contain valuable information for pathogen identification, antimicrobial resistance probability, and virulence risks, for microbial and antimicrobial resistance diagnostics. First, biomarker signatures emitted by clinically relevant bacteria species (e.g., Escherichia coli, Pseudomonas aeruginosa, and Staphylococcus aureus) and the role of sampling are discussed. Then, the importance of machine learning is emphasized for the interpretation of complex volatile signatures, and the recognition of subtle variations in emission profiles to generate diagnostic information. Molecular gas sensors based on chemo/bio-responsive materials are emerging as core technology for affordable and rapid diagnostic tests. To account for the diverse nature of species in the bacterial volatilome, various sensor types, material structures, and modifications of the surface-biomarker interface are analyzed. Finally, we elaborate on translational aspects, including seamless integration into diagnostic microbiological workflows and compatibility with the diagnostic toolkits of general practitioners.