Antioxidants in fruit and vegetable juices have become increasingly popular because of their potential health benefits. Nowadays, juice mixes made from berries present frequent consumer choices, due to their nutritive value and high content of bioactive compounds. Commercial fruit and vegetable juices available in Serbian markets (n = 32) were analyzed for the physicochemical properties, chemical composition, and antioxidant activity. Relative antioxidant capacity index was used for the ranking of the juices according to antioxidant capacity, while antioxidant effectiveness of phenolic compounds contained in juice samples was investigated depending on phenolic antioxidant coefficients. Principal component analysis was applied to study the data structure. In addition, a multi-layer perceptron model was used for modeling an artificial neural network model (ANN) for prediction antioxidant activity (DPPH, reducing power, and ABTS) based on total phenolic, total pigments, and vitamin C content. The obtained ANN showed good prediction capabilities (the r2 values during training cycle for output variables were 0.942). Phenolic, pigments, and vitamin C contents showed a positive correlation with the investigated antioxidant activity. The consumption of commercial berry fruit juices available in Serbian markets may deliver great health benefits through the supply of natural antioxidants.
Keywords: Antioxidant activity; artificial neural network model; multi-layer perceptron model; principal component analysis; relative antioxidant capacity index.