Deciphering the key sweetness-driving compounds in sauce-flavor Baijiu by combining sensomics and multivariate statistical analysis

Sweetness is a key sensory attribute of sauce-flavor Baijiu (SFB), yet its chemical basis remains unclear. Sensomics and multivariate statistics were integrated to decipher sweetness-driving compounds in SFB. Forty-one samples were classified into high-, medium-, and low-sweet groups based on sweetness intensity. Five sweet-tasting polyols and sugars were identified through concentration comparison, Pearson correlation, and partial least squares regression (PLSR). Additionally, 12 sweet-aroma compounds were screened using a combination of PLSR, random forest, and SHAP, capturing both linear and non-linear contributions to sweetness. Recombination experiments of the 17 compounds successfully reproduced the sweetness profile of SFB, and omission tests confirmed 10 compounds as primary sweetness contributors, including D-glucose, D-glucitol, ethyl acetate, ethyl valerate, ethyl heptanoate, ethyl caprylate, ethyl hexanoate, hexyl isobutyrate, 2-phenethyl alcohol, and furfuryl alcohol. This study clarifies the chemical basis of sweetness perception and establishes an interpretable strategy for flavor optimization in complex SFB systems.