Mechanisms of volatile flavor formation in fermented meat products and prospects of artificial intelligence-based prediction

Volatile flavor formation in fermented meat products is governed by complex interactions among raw materials, microbial metabolism, proteolysis, lipid oxidation, amino acid conversion, processing conditions, and ripening. These processes generate diverse volatile compounds, including aldehydes, ketones, alcohols, acids, esters, and sulfur- and nitrogen-containing compounds, which collectively determine product aroma and sensory quality. However, their dynamic, nonlinear, and multi-factorial nature makes accurate prediction and mechanism interpretation challenging using conventional statistical approaches alone. This review summarizes the major volatile compounds, formation pathways, and key factors affecting flavor development in fermented meat products. It further discusses the data foundations required for artificial intelligence-based modeling, including physicochemical indices, process parameters, volatile profiles, electronic nose and GC-IMS fingerprints, sensory evaluation, microbiome, and multi-omics data. The potential applications of machine learning, deep learning, time-series modeling, multi-omics integration, and explainable AI are highlighted for flavor prediction, maturity identification, key factor screening, sensory perception prediction, and mechanism analysis. Future studies should focus on standardized databases, external validation, and interpretable, transferable multimodal models to support precise flavor regulation and intelligent quality control in fermented meat products.