Know more

Context and objectives

The reformulation of foods to meet nutritional recommendations (reduced salt, fat and sugar content) is a major concern for food manufacturers. However, few studies to date have coupled the nutritional and sensory impacts associated with these food reformulations, and even fewer have also taken into account the environmental impact associated with these reformulations. As of yet, no decision-making tool exists to support the reformulation of foods to meet set nutritional recommendations with acceptable sensory qualities, and taking environmental impact into account.

The objective of the NutriSensAl project is to build a decision-making tool informed by an ontology that links and combines data and expert knowledge from these different fields. The originality of the proposed approach lies in the structuring and unified formalization of the knowledge generated in different projects spanning complementary fields with different objectives, making it possible to articulate these different sources of knowledge and mobilize them together, and for new knowledge to be identified.

Project content

NutriSensAl brings together scientists with complementary skills in computer science, process engineering, sensory analysis, nutritional criterion evaluation, and eco-design. Among other things, the nutritional and sensory part of the project relies on data from the following completed and ongoing projects: FP7-Terifiq, FP6-TrueFood, ANR-SensInMouth, ANR-AlimaSSens, and Qualiment-PATATE. The part of the project relating to processes and eco-design draws on data from the projects Qualiment-CellExtraDry, FP7-CAFE, and FP7-SmartRipe.

Preliminary work was carried out to develop an initial PO² ontology based on data on both the eco-design of stabilized micro-organism production processes and the reformulation of dairy products to meet nutritional and sensory criteria. An initial BaGaTel database was structured on the basis this ontology.

The innovativeness of the project lies in the creation of a prototype decision-making tool. This prototype, guided by the PO² ontology, uses data from the BaGaTel database, processing them with probabilistic relational models to identify relations between data from different projects, retrace the context in which they were obtained, and answer initial questions that would otherwise have been unresolvable today, such as:

• For a given variation in the composition of a milk gel, what is the corresponding variation in the perception or digestibility of the product? Do the destructuring parameters at different levels of the digestive tract explain these variations?
• Using knowledge of the destructuring of food in the mouth based on the physiology of the subjects, what food reformulations can be envisaged for specific populations (e.g. elderly people, young children)?
• Which food structure provides a bowl structure compatible with the digestate structures previously defined as optimal for good protein digestibility?
• In order to achieve high environmental performance, which eco-design levers (choice of raw materials used, process management, etc.) should be mobilized and how?
• How can environmental performance be coupled with nutritional and sensory performance? What compromises ought to be made?

One of the original features of the NutriSensAl project is therefore the ability to predict effects that had not been considered in the different projects carried out independently and thus to discover new knowledge.

Industry club

The NutriSensAl project also includes an industry club designed to provide a proof of concept that the tools developed can be enriched and used by industrial partners.

The areas of work within the industry club are:

• the conditions that would allow industrial actors to add their own datasets to BaGaTel;
• ways of managing confidential data
• the possibility of querying BaGaTel on an extended private/public dataset
• the suitability of the BaGaTel interface to meet data entry and query needs