AI is supposed to help with the diseases of civilization

This applies to heart disease, diabetes, obesity and cancer. Metabolic changes are responsible for a number of serious diseases – and they are caused by what we eat. But each person’s diet has a slightly different effect, and until now it has not been possible to ascertain exactly how much the eating habits of specific people affect their health. This is set to change with new research by European scientists, which will study the relationship between food and the body – at the level of a group of individuals and with the help of several technologies, including artificial intelligence. Experts from the Anti-Terrorism Unit will contribute to it.

At the same time, technology is popping up everywhere. For example, the principle by which scientists will discover what a person is actually eating is hidden in a tiny, barely noticeable camera worn on the ear: the device uses computer vision and deep learning, so it recognizes the type of food and the approximate size of portions. Other techniques analyze the gut microbiome or metabolites in the urine. Together, scientists can understand how the body processes food.

The project, called CoDiet, is led by the Spanish research center AZTI, is supported by the prestigious Horizon Europe program and will test a new approach to disease prevention. Its goal is to develop an AI-based tool that can assess individual risks of diet-induced diseases and provide user-tailored nutritional advice.

“Everyone’s metabolic response to the same diet is known to vary. CoDiet will personalize nutritional advice rather than a ‘one size fits all’ approach,” said research leader Itziar Tuerosová of AZTI. Along with her and other experts from a number of fields, experts from FEL of the Czech Technical University are working on learning methods that can link different data.

“Does a particular hormone affect what we crave, or does our diet affect the concentration of a particular hormone? This is the question for which we are looking for an appropriate answer at the level of algorithms. Machine learning revealing causation is a big open problem in both statistics and artificial intelligence, and our methods aim to New based improvement to help solve it. explains Jakob Marijic of FEL of the Czech Technical University.

According to the World Health Organization, serious noncommunicable diseases kill 41 million people each year, accounting for 74 percent of all deaths worldwide. But its mechanisms are still largely unknown. Also because the research that focuses on them relies on information being entered by the users themselves, which can be unreliable. Additionally, it does not focus on vulnerable groups, such as people from lower socioeconomic backgrounds. However, these diseases are often over-represented.

Seventeen institutions will participate in the aforementioned scientific project, in addition to experts from the Czech Republic and Spain, organizations from, for example, Great Britain, Israel and Belgium.