The most promising way to reduce the carbon footprint of pig meat and to increase the competitiveness of our production system is by improving feed efficiency and making rational choices regarding feed ingredients. However, the selection of feed ingredients with a low ecological footprint must be carefully balanced against potential effects on production efficiency and feed costs, to ensure that overall improvements in carbon footprint remain achievable and economically viable. Additionally, further efficiency gains can be achieved by capturing changes in nutrient requirements over time (e.g., farm‑specific seasonal trends), allowing feed composition to be tailored to those needs. NutriTwin is a Flemish 4-year cSBO project (December 2025-December 2029) that at its core investigates how dietary nutrient flows on a pig farm can be modelled into a dynamic digital representation of the production process (i.e. a digital twin). By continuously feeding up-to-date, farm-specific training data into a generic digital twin architecture, the system evolves into a digital twin tailored to that specific farm, accurately capturing its respective nutrient requirements and utilization efficiencies. At the same time, this project also addresses the complexity of real-world digital twinning by exploring the applicability of federated learning and Solid data vaults. As data-ownership and data-sensitivity are important issues to most chain actors (e.g. feed producers, farmers and slaughterhouses), these methodologies could prove an elegant solution, allowing for decentralized updating of the digital twin models without compromising the chain actor's own data to other partners. Finally, this digital twin framework will be employed to optimize feed costs and animal production efficiency (i.e. multi-objective optimization). Secondly, the digital twin also allows to transform the traditional static LCA methodology into a dynamic LCA methodology that enables the capture of real-time environmental impact variations due to specific feeding decisions. This project is a cooperation between UGent (LANUPRO & Biovism Lab groups), ILVO T&V (Data & LCA groups), ILVO Dier and Flanders’ FOOD. Additionally, an Industrial Advisory Board (including different local slaughterhouses, feed producers, integrators and farmers) is also included in the project to provide input and relevant data.

Source: Annual AgData Use Case Summit 2026 booklet.