Modeling of the phenolic compounds and antioxidant activity of blueberries by artificial neural networks for data mining

Abstract

The present work’s goal was to evaluate the effect of different production and conservation conditions, as well as extraction procedures on the phenolic compounds and antioxidant activity of blueberries from cultivar Bluecrop. The production factors considered were origin, altitude of the farm location and age of the bushes, and the conservation was under freezing as opposed to the fresh product. The extraction procedures included two different solvents and different orders of the extraction. Data analysis was performed by training artificial neural networks to model the data and extract information from the model. The results showed that the type of extract and the order of extraction influence the concentrations of phenolic compounds as well as the antioxidant activity of the different samples studied. As to the origin of the farms from where the blueberries were collected, it was found to significantly influence these properties, so that the blueberries from Oliveira do Hospital showed less phenolic compounds and lower antioxidant activity. Older bushes at higher altitudes seem to produce richer berries. Regarding conservation, no influence was observed for phenols but a slight influence could be detected for antioxidant activity.