Overcoming Food Safety Challenges via Mid-Infrared Spectroscopy and Multivariate Data Analysis

Food safety is increasingly challenged by both anthropogenic and natural factors. Among the former, intensive agricultural practices have led to the widespread use of pesticides. While effective for crop protection, pesticides are toxic to humans at high concentrations, and even chronic low-level exposure has been linked to disruptions of biological processes, oxidative stress, and immunotoxic effects. Natural threats include mycotoxins, toxic secondary metabolites produced by fungi. Mycotoxins can cause anorexia, vomiting, reduced nutrient absorption, and immune suppression in both humans and animals; several are classified as carcinogenic.
In this study, we demonstrate how mid-infrared (MIR) spectroscopy combined with multivariate data analysis can enhance food safety by enabling rapid and reliable detection of contaminants. Mid-infrared spectroscopy facilitates the development of portable devices suitable for on-site monitoring, while its integration with chemometric methods strengthens analytical performance. Unsupervised techniques such as Principal Component Analysis (PCA) support pattern recognition and outlier detection, whereas supervised methods, including Sparse Partial Least Squares Discriminant Analysis (SPLS-DA) and Random Forest (RF), enable classification of contaminated versus uncontaminated samples and, in some cases, quantitative prediction of contaminant levels via Partial Least Squares Regression (PLSR). Wheat was chosen as the model commodity, as it provides approximately one-fifth of global caloric intake and remains the principal staple food in many regions worldwide.