FAIR chemosensory data: Unlocking AI for flavour, food and health

Despite its profound relevance to food choice and consumption, nutrition, food safety and public health chemosensory science lags other scientific fields in data sharing, standardization and infrastructure. This gap has real consequences: lack of (shared) data, as well as fragmented datasets, prevent cumulative progress, hinder cross-organizational validation, and limit the development of predictive models that could revolutionize food and consumer science, personalized nutrition, clinical diagnostics and beyond. Groundbreaking scientific breakthroughs emerged in recent years through the combination of advancements in artificial intelligence algorithms and access to large-scale, high-quality and well-organized data. Without this foundation, the chemosensory field risks falling further behind, unable to leverage the transformative potential of AI that is revolutionizing protein structure prediction, drug-discovery, genomic interpretation, medical diagnostics, and other domains. In this commentary we highlight inspiring examples from other fields, acknowledge the state of the art in chemosensory data organization efforts, and outline steps towards the Findable, Accessible, Interoperable, and Reusable (FAIR) principles adoption in chemosensory science. Finally, we envision federated learning scenarios suitable for confidentiality-preserving collaboration with industry and extend an invitation to share raw data from already published academic studies via the “ChemoSensoryData” community in Zenodo.