Foodborne bacteria are widespread contaminated sources of food; hence, the real-time monitoring of pathogenic bacteria in food production is important for the food industry. In this study, a novel rapid detection method based on microbial volatile organic compounds (MVOCs) emitted from foodborne bacteria was established by using ultraviolet photoionization time-of-flight mass spectrometry (UVP-TOF-MS). The results showed obvious differences of MVOCs among the five species of bacteria, and the characteristic MVOCs for each bacterium were selected by a feature selection algorithm. Online monitoring of MVOCs during bacterial growth displayed distinct metabolomic patterns of the five species. MVOCs were most abundant and varied among species during the logarithmic phase. Finally, MVOC production by bacteria in different food matrixes was explored. The machine learning models for bacteria cultured in different matrixes showed a good classification performance for the five species with an accuracy of over 0.95. This work based on MVOC analysis by online UVP-TOF-MS achieved effective rapid detection of bacteria and showed its great application potential in the food industry for bacterial monitoring.
Keywords: MVOCs; foodborne bacteria; machine learning; mass spectrometry.