In recent years, the integration of spectroscopic techniques with machine learning algorithms has emerged as a powerful analytical paradigm, demonstrating significant potential across diverse sectors, including food safety, pharmaceuticals, and agriculture. This review focuses on the application of this combined approach in food safety analysis. It begins by outlining the fundamental principles of common spectroscopic methods and the theoretical foundations of relevant machine learning models. The core of the discussion systematically examines their specific applications in three critical areas: the identification of food adulteration, qualitative classification of unsafe components, and quantitative detection of contaminants or residues. By synthesizing recent advances, this study aims to provide a clear overview of the methodology's capabilities. Furthermore, it concludes by critically addressing the key technical and practical challenges that currently limit widespread industrial adoption, thereby highlighting pathways to translate this promising technology from laboratory research into robust real-world solutions for enhanced food safety monitoring.
Ma et al. (Mon,) studied this question.