Chemometric non-targeted analysis for detection of soybean meal adulteration by near infrared spectroscopy

In this study a quick and efficient routine procedure for food fraud detection by multiple adulterants is presented. Non-targeted analysis employs the Near Infrared (NIR) spectroscopy measurements and one-class classification modeling as the chemometric data processing. The approach is illustrated by the analysis of a large collection of NIR spectra of soybean meal. The clean and contaminated samples are studied. The main advantage of the proposed approach is that it is not aimed at identification and quantification of a specific contaminant. The procedure is designed in such a way that it detects any deviations from the clean samples. The non-targeted analysis has its own limit of detection (LoD). In the study we have presented an approach for LoD assessment. This issue is of great importance for practical applications. The proposed approach can be applied for other types of feed and food products.

Automated summary

This study presents a quick and efficient routine procedure for food fraud detection using NIR spectroscopy measurements and one-class classification modeling for non-targeted analysis. The main advantage is the detection of deviations from clean samples rather than specific contaminants, with a method for limit of detection assessment provided. The proposed approach has potential applications for other feed and food products.