The potential new microbial hazard monitoring tool in food safety: Integration of metabolomics and artificial intelligence

Abstract

Background

For a sustainable food processing environment, robust and real-time monitoring of pathogens is particularly important. Therefore, novel methods integrating metabolomics and artificial intelligence for early detection, identification, and micro-risk prediction have received significant attention from researchers in recent years. However, the absence of standardized procedures for data acquisition, quality control, and authenticity evaluation still hampers the development of this field. In addition, large datasets necessary for training models to accurately manage controls within food matrices, as well as the lack of any universal model that can be applied across all scenarios, are also challenges that need to be addressed.

Scope and approach

Metabolomics when combined with deep learning (DL) has indicated significant potential in food microbial monitoring. This review covers the reported applications in this area while highlighting early detection of microbial contaminants. Traditional and novel metabolomics have been compared and limitations, challenges, and prospects in this area are discussed. The key focus is discussing the role of DL in improving the application of metabolomics in the classification and identification of foodborne pathogens.

Key findings and conclusions

Some publications in this field have demonstrated the role of metabolomic biomarkers, fingerprints, and profiles in the identification and early detection of microbial risks. The workflow for screening and validating biomarkers of pathogenic microorganisms in food matrices is currently underway. The integration of artificial intelligence (AI) and metabolomic profiling indicates high potential in the real-time monitoring and identification of microbial hazards at various stages of food production, transportation, and consumption.