New research project MetaboTwin to develop a computer model of how our bodies respond to food

A team of scientists from Wageningen University & Research (WUR) and Eindhoven University of Technology (TU/e), and several industrial partners from food and nutrition, health-tech, and (digital) biomarkers sectors are joining forces to build a Human Dietary Digital Twin, an advanced computer model that predicts how individuals respond to real-life meals.

When we eat, our bodies experience short-term changes in blood sugar (glucose), hormone (insulin), and fat (triglyceride) levels, known as postprandial responses. These metabolic responses show how efficiently the body processes nutrients after a meal. Over time, repeated unfavourable responses can increase the risk of various diseases such as type 2 diabetes and cardiovascular disease. Interestingly, every person’s metabolism is unique: research shows that individuals can have very different postprandial responses to the same foods or meals. 

“People respond differently to foods and meals, but we still don’t fully understand why.” says Dr. Anouk Gijbels from Wageningen University & Research. “Furthermore, for instance for glucose responses, it’s not just the amount of carbohydrates that matters: other factors such as fat, protein, fibre, and texture also play an important role. In this project, we aim to untangle these interactions, considering both individual differences and meal characteristics, to understand how different foods affect post-meal metabolism among different people.”

To explore these factors, the MetaboTwin team will conduct highly controlled dietary intervention studies to measure how people respond to complex meals that resemble real-life eating patterns. 

Measuring metabolic responses to different meals in such studies is time-consuming, costly, and demanding for participants. Using clever computer models to better understand and predict these responses would help researchers to explore the health effects of many meals more efficiently and at larger scale.

In earlier collaboration, TU/e and WUR already developed a computer model that describes how blood glucose, insulin, and triglyceride levels change after standardised test meals, basically milkshakes that are high in fat and sugar. MetaboTwin will now extend this model to capture responses to more complex, real-life meals containing different amounts and sources of protein, as well as incorporating additional meal characteristics such as texture and fibre content.

The data from the new human studies will be combined with existing high-quality nutritional datasets to develop the Dietary Digital Twin. This virtual model allows researchers to simulate how an individual’s metabolism responds to different foods and meal combinations without the need for repeated testing in humans.

“Digital twins are widely used in engineering to simulate and optimise systems like aircraft engines or manufacturing processes,” explains Professor Natal van Riel from TU/e. “But the human body is far more complex than any machine. With MetaboTwin, we are combining biology, mathematics, and AI to create a digital representation of human metabolism. This model can help us understand, predict, and improve how our bodies respond to food.”

By creating a computer model of human metabolism, MetaboTwin will help design healthier and more sustainable foods and enable personalised dietary advice tailored to each person’s metabolic profile, supporting the prevention of chronic nutrition-related diseases such as type 2 diabetes and cardiovascular disease.
The project is supported by the Top Consortium for Knowledge and Innovation (TKI) Agri & Food and private partners Thai Union, Herbalife, Norilia, Ani.AI, Nuritas, and Genos.