Raman spectroscopy powered by machine learning methods for rapid identification of foodborne pathogens

Foodborne pathogens pose a serious threat to food safety, and micro-confocal Raman spectroscopy is emerging as a promising method for the rapid detection and identification of them. However, the high degree of similarity in the biochemical composition of different foodborne pathogens, particularly among the serotypes of the same genus, degrades the identification accuracy. To address this challenge, we optimized multiple machine learning methods based on Raman spectroscopy and performed a thorough comparative study on them for the discrimination and prediction of seven types of foodborne pathogens originating from five different genera. The results indicated that the improved clustering algorithms can identify phylogenetic relationships among pathogens, and the designed dual-scale Convolutional Neural Network (CNN) model achieved a superior identification performance, with a prediction accuracy exceeding 98.4%. These optimized machine-learning-driven Raman spectroscopy methods are expected to become a promising tool for the rapid detection of microbial contamination in food.