Fresh-food cold chains are still managed mainly through temperature threshold alarms and static calendar expiry dates. These signals do not show the actual remaining quality of a shipment: two loads with very different temperature histories may receive the same operational treatment, while pricing, write-offs and routing decisions are often made without knowing the real shelf-life left.

Current market solutions cover separate layers only. Visibility platforms capture telemetry but do not predict quality or take decisions; point quality measurements do not follow goods across transport and operators; dynamic markdown tools typically use date and demand rather than measured remaining shelf-life; refrigeration optimisation works within fixed temperature limits. In multi-operator chains, quality data is also fragmented between distributor, carrier and retailer systems, so the quality assessment is often reset at each handover.

The market need is a trusted decision layer that converts existing operational telemetry into usable quality evidence, supports legally and commercially robust decisions with quantified uncertainty, and enables action before food becomes unsellable.

FRESHCHAIN will add a decision layer above existing cold-chain monitoring systems. The platform will identify degradation kinetics per product from sparse, noisy operational telemetry and temperature history, using grey-box and physics-informed models rather than relying on laboratory-only isothermal calibration or manual point measurements.

Predicted remaining shelf-life will be accompanied by conformally calibrated uncertainty intervals, so automated decisions are triggered only when confidence is sufficient and uncertain cases can fall back to human review. This enables a closed loop from sensor data to prediction and then to action: rerouting shipments to suitable customers, applying dynamic markdown based on actual quality, or selling surplus through a marketplace.

The solution will introduce a multi-operator quality passport, built on GS1 EPCIS 2.0 concepts, to carry remaining shelf-life, confidence and model provenance across handovers without exposing raw telemetry. It will also develop quality-aware model predictive control for refrigeration, where cooling effort is adapted to the predicted quality reserve of the goods instead of fixed setpoints.

The consortium will define the fresh-food cold-chain use cases, decision points and quality-data requirements for distributors, transport operators and retailers, using the Slovenian SmartLogiFresh base and GEAPRODUKT distribution pilot as reference environments.

Technical work will cover telemetry ingestion, data cleaning and temperature-history reconstruction; development of SKU-level kinetic quality models; conformal uncertainty calibration; and few-shot transfer methods for onboarding new products and routes with limited labelled data. The partners will validate remaining shelf-life predictions against reference measurements and operational outcomes.

The project will build the decision components for shipment rerouting, quality-based markdown and surplus marketplace triggering, including fail-safe rules when prediction confidence is insufficient. It will also develop a quality-aware cooling optimisation module that balances energy consumption with predicted remaining shelf-life.

Interoperability work will specify and implement a quality-passport format on top of GS1 EPCIS 2.0, including provenance, remaining shelf-life and confidence information, with privacy-preserving exchange of quality judgements rather than raw telemetry. The integrated platform will be piloted in realistic cold-chain operations and assessed against KPIs for prediction accuracy, uncertainty coverage, decision latency, waste reduction potential, data continuity across operators and energy savings.