Comprehensive Flavor Profiling of Dairy Products Using Electronic Tongue: Discrimination Based on Processing Parameters and Formulations.

Flavor is a critical determinant of consumer acceptance for dairy products. However, objectively deconvoluting the contributions of multifactor interactions-such as processing conditions, formulations, and raw material properties-remains challenging. This study employed the INSENT TS-5000Z electronic tongue (E-tongue), supplemented by sensory evaluation and chemical analysis, to profile eight fresh milk and six plain yoghurt samples from five leading brands. Key variables included sterilization methods, milk sources, and sugar content. Nine taste indices were quantified. Principal component analysis (PCA) effectively discriminated samples based on these parameters, with cumulative variances of 81.2% for milk and 92.5% for yoghurt. Organic milk (SY-JZ-ORG) exhibited significantly higher bitterness (10.77 ± 0.04 TU) and astringency (4.60 ± 0.02 TU), which correlated with elevated levels of α-linolenic acid, hydrophobic peptides, and the creamy lactone δ-decalactone. Full-sugar yoghurt (MN-Y) showed pronounced astringency (9.38 ± 0.20 TU), consistent with its highest sensory scores for "thickness" and "astringency." Partial least squares regression (PLSR) models reliably predicted sensory attributes from E-tongue data (R2Y = 0.971-0.972, Q2 = 0.856-0.902), identifying astringency as the primary predictor. This study demonstrates that the E-tongue can objectively decode complex flavor differences arising from multifactor interactions, offering a robust framework for dairy quality control and product development.