Tailored nutrition advice thanks to digital twin
Wageningen scientists use a digital twin to predict how people will respond to meals. Not everyone responds in the same way to a meal rich in sugar or fat—the ultimate goal: a digitally generated tailored diet recommendation.
Eat plenty of vegetables and fruit, avoid sugars and be careful with fats. These are familiar recommendations for a healthy diet. There is nothing wrong with these recommendations, say Wageningen scientists, but they are too generic. Each individual responds differently to nutrition. Suppose two people eat a banana, for example. One may experience a spike in their blood sugar level while the other's blood sugar barely increases. The degree to which this increase manifests affects metabolic diseases such as diabetes. Artificial intelligence (AI) makes personal nutrition advice possible.
Simulating the body's response
The absorption of fat also differs from one person to the next. The lipid content in the blood after eating a hamburger varies per person. This is valuable information because the degree to which the fat content in the blood increases is a key predictor of cardiovascular disease. Unlike the glucose level, there are no sensors that can monitor the blood's lipid content. WUR does, however, have data on over five hundred middle-aged persons with excess weight that were gathered during previous studies. This data is used to build a digital model that can be used in a digital twin, which can simulate and predict the body's reaction.
A digital copy enables you to adjust circumstances and predict the effects, says researcher Diederik Esser. A team of researchers led by Lydia Afman works on a biological digital twin under the name 'Me, my diet and I'. The team consists of nutrition experts like Esser, behavioural scientists, bio-informaticians, engineering business administration experts and consumer researchers.
Personal health advice
The ultimate goal is a personalised diet that includes BMI, age, fat distribution, blood pressure, and nutrition, improving glucose and lipid spikes in the bloodstream. The diet recommendations become increasingly accurate by comparing the recorded blood values to those predicted by the digital twin. This information is then integrated into a personal diet recommendation that takes into account personal preferences, such as taste and an appreciation for organic products. Diets that include personal preferences are more likely to be followed.
Responding to concerns
The team is to test a prototype of the model on human subjects to assess whether the predictions are correct. This will reveal the predictive value. Moreover, the data obtained from test subjects can be used to improve the prototype further, according to engineering business administrator Marc-Jeroen Bogaardt. Within the project, he focuses on data governance, infrastructure and stakeholder commitment throughout the development of the digital twin. This means that the target group, people with health issues, are involved in the project from the start.
Wageningen scientists want to include the end user's wishes in the tool. Some people are wary of the digital twin being managed by a commercial party rather than the university. Others are less concerned. Researchers aim to respond to these and other concerns to increase the use of the tool, and thus its impact.