Food data space-data collection procedures; data management platform and a data lake, both compliant with GDPR. Implemented in three EU case scenarios (Athens, Galicia, Lithuania).
Metrics for food system assesment, integrating nutrition, health, social, economic and environmental dimensions to be displayed through an eco-healthy fingerprint visualization plot.
Analytical methods based on ML to manage heterogeneous data from diverse actors in the food system; reduce computational costs; and provide interpretable and explainable results (xAI).
Decision-making tools:
·Predictive Dashboard: allows administrations, companies and consumers to address their policy-, business- and consumption-choices extending sustainability in the food system with a robust scientific grounding. This approach identifies which actors can influence or direct this transformation.The predictive AI-empowered methods will enable decision makers to underpin their decisions with a solid scientific basis, built from evidence, with an appropriate quantification of the ensuing uncertainty, and in a comprehensible manner.
·Consumer mobile app: integrates AI-based insights, eco-healthy fingerprints, and reformulated diets, continuously improved with user feedback, thus linking technical achievements to societal impact.
The first prototypes have been developed and validated for their suitability to the identified user stories and their fitting to the case studies: predictive analysis of time series data, estimating evolution and interaction of social, political, legal, economic, technological, food and climate drivers that affect sustainability; clustering analysis of food system trajectories, identifying alternative pathways across production, transformation, consumption, and waste; digital twins to simulate the impact of sustainability indicators at midpoint and endpoint levels; causal discovery techniques, linking high-level economic indicators, regulations, and production trends to their environmental and social impacts; xAI approaches, combining expert knowledge with machine learning outputs to ensure transparency, multilingual accessibility, and trustworthiness of results.
These developments respond to needs identified in the field: data exploitation, integration of multimodal datasets, and provision of decision-support tools that are robust and interpretable. By embedding software sustainability principles and ethical guidelines into system design, SOSFOOD ensures that its AI solutions are not only technically advanced but also trustworthy and usable in real-world contexts.
