The association of food ingredients in breakfast cereal products and fumonisins production: risks identification and predictions

Abstract

Breakfast processed products are remarkably at risk of fungal contamination. This research surveyed the fumonisins concentration in different breakfast products and carried out in vitro experiments measuring fumonisins content in different substrates inoculated with Fusarium verticillioides. The pipeline started with the identification of combinations of ingredients for 58 breakfast products. Twenty-three core ingredients, seven nutritional components and production types were analyzed using a Pearson correlation, k-means clustering, and principal component analysis to show that no single factor is responsible for high fumonisins detection in processed cereals products. Consequently, decision tree regression was used as a means of determining and visualizing complex logical interactions between the same factors. We clustered the association of ingredients in low, medium, and high risk of fumonisin detection. The analysis showed that high fumonisins concentration is associated with those products that have high maize concentrations coupled especially with high sodium or rice. In an in vitro experiment, different media were prepared by mixing the ingredients in the proportion found in the first survey and by measuring fumonisins production by Fusarium verticillioides. Results showed that (1) fumonisins production by F. verticillioides is boosted by the synergistic effect of maize and highly ready carbohydrate content such as white flour; (2) a combination of maize > 26% (w/w), rice > 2.5% (w/w), and NaCl > 2.2% (w/w) led to high fumonisins production, while mono-ingredient products were more protective against fumonisins production. The observations in the in vitro experiments appeared to align with the decision tree model that an increase in ingredient complexity can lead to fumonisins production by Fusarium. However, more research is urgently needed to develop the area of predictive mycology based on the association of processing, ingredients, fungal development, and mycotoxins production.

Machine Learning Applied to the Detection of Mycotoxin in Food: A Systematic Review

“…In a study related to identifying and predicting risks related to the presence of fumonisins in breakfast cereal products, Ref. [ 100 ] developed a model specifically designed to predict the risk of fumonisin contamination, with a particular emphasis on a mixture of ingredients. In their research, fifty-eight distinct breakfast products were purchased from local grocery stores in Florence, Italy, during 2019.…”