Chemical Drivers of Sweetness, Sourness, and Astringency in Apple Juice Identified Using Support Vector Machine Regression.

The role of the main apple compositional parameters (sugars, organic acids, polyphenols, and volatile organic compounds) and their interactions on the perceived sensory attributes of Sweetness, Sourness, and Astringency was investigated in apple juice. Bivariate correlations and Support Vector Machine Regression (SVMR) analyses were used to examine the relationships between all the analyzed features and their relative importance in an interactive mode. By analyzing the contribution of the features to the SVMR models using Shapley values, we found that sweetness was strongly influenced by sucrose and by a few volatile organic compounds (VOCs), including acetate and butyl esters associated with fruity odors and flavors. For sourness, as expected, a strong contribution is given by malic acid, while VOCs associated with fruity odor contribute negatively. Regarding astringency, the major contributions were given by malic acid, procyanidins, sorbitol, and chlorogenic acid; the VOCs (few esters, hexanol, and (E)-2-hexenal) show a suppressing effect on astringency. In our investigation, SVMR confirms the role and the importance of the evidenced compositional parameters in driving the perception of the three analyzed sensory attributes in apple juices. PRACTICAL APPLICATIONS: This research could help provide apple breeders with new varieties with more predictable flavor, such as sweeter apples for juice or more tannic apples for cider. It could also help juice and cider producers select, blend, and process apples more consistently by using chemical measurements to estimate how a product may taste. The work may support better quality control, reduces variability, and strengthen product differentiation in markets where flavor consistency and clean-label formulations are increasingly important.