Project

Call for Partners | SMART: Modelling sensory perception through hybrid simulation of food processing and formulation

Food manufacturers and ingredient suppliers face rising pressure to deliver healthier, more sustainable, and affordable products without compromising taste or quality. Reducing salt, sugar, fat, and animal-derived or unsustainable ingredients is challenging, as it can affect flavour, texture, shelf life, and consumer appeal. Reformulation also requires navigating complex regulations and rethinking sourcing and processing.

To address these challenges, we are launching SMART, a collaborative consortium bringing together food manufacturers, ingredient suppliers, and researchers to accelerate reformulation using AI, enabling faster, smarter innovation and more effective product development. 

The SMART approach  

SMART develops hybrid models combining classical mechanistic models and modern machine learning, making use of established expertise and cutting-edge technology. This enables to predict how ingredient and process changes will affect the final product. This hybrid AI approach has several benefits, the most important one being that you get more value from the same data or need less value for achieving the same value. Further knowledge integration leads to scientifically sound models. These models will integrate processing conditions and formulation parameters to predict sensory perception. This approach allows SMART’s partners to uncover hidden relationships, including cross modal interactions, between formulation choices and key control parameters.  

Within the project, we will focus on a single model product as a demo case to generate a high-quality dataset. This dataset, which may be enriched with publicly available data, will serve as the foundation for developing a flexible and robust predictive model. A key objective is to establish a systematic, high-throughput data generation approach that supports the creation of reliable models and enables future integration of partner-specific data. This structured methodology is a crucial milestone within the project. 

The demo case will also serve to demonstrate the model’s feasibility and real-world value in a practical setting. Although the scope of the project is limited to this one showcase application, the resulting model will be designed for broader use. After the project, partners will be able to apply the model to their own products and processes by integrating their own data. This will support targeted product-specific optimization in line with their innovation goals. 

Summarizing 

SMART integrates data on processing, food physics, flavour and sensory analyses, and machine learning to:

  • Predict how ingredient and process changes affect product quality
  • Reduce reformulation time and cost
  • Improve health, sustainability, and consumer appeal of food products

Partners 

We invite food manufacturers, ingredient suppliers, and food processing technology providers to join the SMART project submission for the TKI Agri & Food PPS grant call, the deadline of the submission is September 1st, 2025. 

Partners will actively invest in the project through an in-cash and in-kind contribution, demonstrating joint commitment to accelerating innovation in food production and (re)formulation.