Aroma-Enhancing Strategies in Mead: A Metabolomics- and Machine Learning-Guided Additive Approach

Abstract To enhance dry mead (DM) flavor, 13 herbal teas and coffee were added, and volatile changes were dynamically monitored using HS-SPME-GC-MS with three columns. While aroma intensity improved, off-flavors also increased. Pearson correlation analysis identified 17 key volatile compounds associated with enhanced aroma and reduced off-flavor intensity. The Random Forest (RF) algorithm was applied, enabling the identification of compounds positively associated with sensory evaluation scores, including both taste and aroma. Four compounds—ethyl phenylacetate (ES30), decanal (AD10), ethyl nonanoate (ES35), and phenylacetaldehyde (AD6)—were selected. Their addition increased aroma-active compounds by 3.5–39.0%, reduced isoamyl alcohol (AL4) by 9.7–56.81%, and raised total ester content by 3.05–5.88 times. Combined with sensory and metabolomic data, single-additive meads better preserved original flavor while enhancing floral, fruity, sweet, and herbal notes and suppressing off-flavors, compared to tea- or coffee-blended meads. Notably, inhibiting isoamyl acetate (ES10) formation appears critical for reducing AL4.