A review of the food authentication research using class-modelling approaches

Food authentication, the process of verifying the alignment between product characteristics and claims, is a critical aspect of the food sector. Besides the analytical technique used to determine food authenticity, a proper and meaningful data processing with multivariate statistics and chemometrics is fundamental. For instance, discriminant and class-modelling approaches are often used in the context of food authentication and both assign objects to predefined classes. However, in scenarios where the main focus is on the 'authentic' class, class-modelling approaches like SIMCA (Soft Independent Modelling of Class Analogy) are preferred. These approaches construct a function for the target class, outlining a region where its samples are more likely to be situated. SIMCA, emphasizing individualized class characterization, yields robust results even when modifications are introduced to the dataset after the model is built. Therefore, the aim of this article is to review the advancements of food authentication using SIMCA. This review is based on a literature search conducted using the Scopus database, considering works published over a 10-year range. A total of 71 research papers have been selected. The review is structured around three key elements: i) the aims of the research, ii) the technologies employed for food authentication, and iii) the specific food products under investigation. Adulterant detection emerged as the most extensively studied issue, with 29 out of 71 articles, highlighting the significance of combating adulteration practices in the food industry. Other key topics include the verification of geographical origin (19 articles), authentication of a food category (17 articles), and confirmation of the species used in products (9 articles). Analytical methods used in these studies revealed the prevalence of spectroscopic analyses, primarily based on NIR and IR spectroscopy (29 articles), Raman, NMR, and UV-Vis spectroscopy (18 articles). The advantages of these methods include the possibility to perform the analysis using portable instruments. Additionally, 19 articles described the application of analytical methods, including chromatography and mass spectrometry. Among food, the categories ‘spices and herbs’ and ‘oil and fats’ were the most investigated. In fact, these classes are identified as particularly susceptible to adulteration and food frauds. SIMCA provides robust results in food authentication even in the presence of dataset modifications. While recent advancements in portable spectroscopic devices offer interesting potentialities, limited studies indicate a need for further exploration in this area. As future perspectives, the review suggests a need to shift attention towards emerging food products, such as insects, meat analogues, and food based on alternative proteins. The dynamic nature of the food supply chain requires continuous adaptation, and future research should address the authentication needs of these innovative products.