From machine learning to neuroimaging: a comprehensive review of flavor prediction and recognition techniques

The study of flavor is undergoing a paradigm shift, moving beyond traditional descriptive approaches towards a predictive and mechanistic science. While previous reviews have often focused on singular technological advances, this review provides a unique integrative perspective by synthesizing a multi-scale research framework that spans from molecular-level prediction to central nervous system interpretation. This review first delineates how artificial intelligence and computational simulations are revolutionizing the in-silico prediction of flavor compounds through structure-activity and ligand-receptor modeling. Simultaneously, it elaborates on how biosensors, built upon functional olfactory/taste receptors, provide unprecedented objectivity for flavor recognition at the biomimetic interface. Crucially, the indispensable role of neuroimaging techniques (EEG, fMRI) is emphasized in directly decoding the brain's integrated perceptual response to flavor-the ultimate arbiter of sensory experience. By synthesizing these advances into a cohesive "from molecule to brain" pipeline, this review not only serves as a roadmap for the future of flavor science but also proposes a unified framework for the field, highlighting the critical role of this integrated approach in guiding the rational design of food products and driving industry innovation.