A machine learning approach to predict texture and sensory attributes of beef meatballs treated with ultrasound and tea polyphenols

The oxidative deterioration of processed meat products during storage remains a critical challenge in the food industry. Tea polyphenols (TP), as potent natural antioxidants, offer a solution for stability enhancement, though their application requires balancing functional efficacy with sensory constraints. This study investigated the effects of ultrasound-assisted TP incorporation (solutions of 0, 2, 4, and 6 g/100 mL, w/v; designated K0–K4) on the physicochemical stability, volatile fingerprint, and sensory profile of boiled beef meatballs. Results indicated that the K3 treatment (4 g/100 mL TP) significantly improved oxidative stability, reducing TBARs and peroxide values by approximately 33% and 22%, respectively, compared to the control. GC–IMS and E-nose analyses revealed that TP treatment reshaped the volatile fingerprint, characterized by the suppression of oxidation-derived aldehydes and sulfur-containing volatiles, alongside the enrichment of specific alcohols and esters (e.g., trans -3-hexenol and ethyl acetate). While sensory profiling identified dose-dependent astringency and textural modifications intrinsic to phenolic incorporation, SHAP-based interpretation of an Extra Trees model highlighted TBARs, moisture content, and sensory scores as the primary determinants of comprehensive quality. Among the evaluated levels, K3 emerged as a robust functional formulation, offering superior oxidation inhibition and structural stability. These findings suggest that ultrasound-assisted TP incorporation is a promising strategy for developing oxidation-resistant meat products, providing data-driven insights for future formulation optimization.