Less than one hour to find contaminants in fresh produce

The GRACED Project overcomes traditional testing limits with fast, portable, and AI-powered decision support

Fresh produce can carry invisible threats. Bacteria, viruses, and chemical residues often go undetected until it’s too late, when products are already on store shelves. Despite significant progress in traceability and logistics, food safety inspections remain largely dependent on random sampling and laboratory-based methods. These methods are accurate but time-consuming and costly, often requiring several days to deliver results. Consequently, only a limited number of samples are tested, leaving many production batches unchecked. This gap in routine quality control creates a persistent blind spot, increasing the risk of microbiological and chemical contamination throughout the supply chain.

To address this critical weakness, the European research and innovation project GRACED aims to transform food safety monitoring. It will introduce a fully integrated sensing and analytics platform designed for speed, accuracy, and operational scalability. In particular, the project integrates advanced biosensing technologies, artificial intelligence, and Internet of Things (IoT) systems to enable real-time, in-field detection of contaminants, thereby eliminating delays and reducing reliance on centralized laboratory infrastructure.

A European response to food safety challenges

GRACED is a European research and innovation project that was completed in December 2024 with €4,989,480 in funding under the Horizon 2020 programme. The project was coordinated by the Cyprus Research and Innovation Center (CyRIC). The Center led the system’s technical development, the integration of the sampling module, the IoT platform, and strategic oversight. Around CyRIC, a multidisciplinary consortium of 17 partners was formed, including public research centers, universities, SMEs, and direct players in the agri-food chain. Among the Italian participants, the National Research Council (CNR) contributed to the optical design of the sensor and spectral data processing. This work was done alongside Tecnoalimenti S.C.p.A., a non-profit organization serving as an interface with industrial and operational needs in the agri-food sector.

The production side was represented by partners like PILZE-Nagy Kft (Hungary), one of Central Europe’s leading mushroom producers using organic substrates (wheat straw is the substrate at risk for mycotoxins). It also included French entities such as Pour Une Agriculture du Vivant, Ver de Terre Production, and Sous Les Fraises. They tested the system in agroecological, urban, and short-supply contexts. Scientific and photonic support came from key institutions such as CNRS and Université Bourgogne Franche-Comté (France), along with the Belgian research center MULTITEL, active in optical development and microfabrication. Other essential contributions came from EGM (Easy Global Market), AMO GmbH (Germany), Bialoom Ltd, Lumensia Sensors S.L. (Spain), and Aristotle University of Thessaloniki (Greece), which worked on the biochemical module and its validation. 

A holistic solution for food quality monitoring

The project’s full title—Ultra-compact, low-cost plasmo-photonic bimodal multiplexing sensor platforms as part of a holistic solution for food quality monitoring—encapsulates its goal: to radically improve food safety control along the fresh produce supply chain. Rather than relying on delayed and selective lab testing, GRACED sets out to develop a field-deployable platform capable of detecting multiple contaminants directly at the point of risk: in farms, processing facilities, or distribution channels.

Detecting contaminants before they become a threat

At the heart of the GRACED platform is a new class of plasmo-photonic bimodal biosensors, engineered to detect a wide range of contaminants directly within liquid samples. Unlike conventional sensors, the GRACED chip combines two complementary signal amplification mechanisms. The first is interferometry, which tracks changes in light phase. The second is plasmonics, which enhances sensitivity through surface interactions at the nanoscale. These two modes work in tandem on a miniaturized photonic circuit, enabling the simultaneous detection of up to seven different analytes, such as pathogens, toxins, and chemical residues.

One of the project’s most important achievements is the development of a polymer-based bimodal interferometric sensor with exceptionally high sensitivity. Laboratory tests confirmed its ability to detect even minimal changes in a sample’s refractive index. Specifically, this makes it possible to identify contaminants at trace levels, well below the detection thresholds of conventional technologies. More importantly, this responsiveness enables GRACED to catch potential hazards early, long before they pose health risks or lead to costly recalls. 

A novel cladding material is the key innovation behind this performance. It keeps optical signals stable, resists chemical interference, and integrates seamlessly into microfluidic systems. This compatibility enables the sensor’s incorporation into compact, automated devices. Devices capable of drawing, processing, and testing liquid samples without manual intervention. It enables faster and cleaner operations. But also supports real-time analysis directly at the point of need, and paves the way for low-cost, portable tools that can be deployed widely across the food supply chain.

Two configurations for diverse monitoring needs

To ensure adaptability across different operational settings, the GRACED sensor was embedded into two distinct device configurations. The first is a portable instrument designed for laboratory and field use, capable of analysing both solid and liquid samples. This device targets professionals involved in routine quality control, enabling mobile inspections in processing facilities, transportation hubs, and distribution centres.

The second configuration, which represents the fixed-installation version of the system, is an autonomous sensing node. It is developed specifically for continuous, unattended monitoring of water-based environments. These include irrigation systems, washing flows, and indoor or vertical farms. These contexts require continuous monitoring but have limited human presence. Designed to operate with minimal maintenance, this node can detect contamination events as they happen, supporting preventive responses and improving traceability.

Smart analytics architecture for predictive interventions

In addition to advanced sensing hardware, the GRACED platform integrates a structured data analytics architecture designed to support timely and informed interventions. At its core is a smart Decision Support System (sDSS). This system interprets biosensor outputs in relation to contextual parameters, including temperature, humidity, and location-specific variables. The system ingests data from GRACED devices and external sources, always in compliance with GDPR and user-specific requirements.

The sDSS is built on a graph-based data structure that captures relationships between various data points. That enables the use of Graph ML, an advanced machine learning techniques. This structure supports three layers of analytics: descriptive (what is happening), predictive (what is likely to happen), and prescriptive (what actions should be taken). This architecture enables the system to issue real-time alerts and actionable recommendations. It effectively transforms raw sensor data into intelligent tools for decision-making. Beyond contaminant detection, the GRACED platform also supports critical functions like traceability and automated documentation. This enables operators to track a product’s origin and handling throughout the entire supply chain. Moreover, it ensures transparency and accountability at every stage, from primary production to final distribution, making GRACED a truly end-to-end solution. In addiction, GRACED integrates smoothly with existing monitoring systems, enabling food businesses to enhance their workflows without overhauling infrastructure.

Validation, scalability, and readiness for adoption

GRACED technologies underwent extensive validation, both in the laboratory and through deployment in four real-world operational environments across Europe. These pilots encompassed a diverse range of use cases. They included conventional outdoor farming, urban agriculture integrated with on-site restaurant services, short supply chains for fresh produce, and automated mushroom production. Across all testbeds, the platform consistently demonstrated its ability to detect multiple types of contaminants, biological and chemical, in under one hour, compared to the several days typically required by standard laboratory methods. 

The system showed high robustness and adaptability, effectively handling various sample types, including leafy greens, irrigation water, and washing flows. Project partners implemented all pilot programs in close collaboration with end users, including farmers and food industry operators.

The industrial scalability of GRACED technologies was also addressed. A second sensor configuration based on silicon nitride and aluminum, materials widely used in the semiconductor industry, demonstrated excellent thermal stability and full compatibility with standard chip fabrication processes. This capability allows manufacturers to produce GRACED sensors as disposable, single-use components. This feature is essential for hygiene-critical applications like fresh food inspection and in-field monitoring.

Commercial potential and open access strategy

Bialoom Ltd has filed a European patent for the GRACED sensing platform to support post-project exploitation, while also sharing many project results on Zenodo and other open-access platforms to ensure accessibility for researchers and industry. Together, these outcomes demonstrate the technical readiness and commercial potential of the GRACED system as a strong example of integration between advanced photonics, microfluidics, and digital intelligence

References

• Berini, A., Berneschi, S., Cosci, A., Dardano, P., Giannetti, A., Gulisano, A., Lauria, A., Margiotta, N., Mazzola, M., Melchionna, D., Neri, G., & Trono, C. (2024).
  SU-8 based plasmonic bimodal interferometers for high-sensitivity sensing. Sensors and Actuators Reports, 6, 100187. 
• Berini, A., Cosci, A., Giannetti, A., Margiotta, N., Trono, C., Dardano, P., & Berneschi, S. (2023).
  Polymer-coated SU-8 waveguides for interferometric biosensing in aqueous media. Optics and Laser Technology, 164, 109498. 
• Dardano, P., Berneschi, S., Trono, C., & Giannetti, A. (2023).
  On-chip interferometric SU-8 sensors for bimodal plasmonic detection of contaminants. Journal of Lightwave Technology, 41(15), 5179–5185. 
• Berini, A., Dardano, P., Berneschi, S., Cosci, A., Trono, C., Margiotta, N., & Giannetti, A. (2025, March).
  SU-8 bimodal plasmonic interferometric sensors for real-time detection of agro-contaminants in aqueous environments. Sensors and Actuators B: Chemical, 407, 134589.