Professor Ilias Tagkopoulos, Director of the USDA AI Institute for Next-Generation Food Systems (AIFS), says we only know about 5 to 10% of the chemicals found in our food. Even less is known about the effects of food ingredients on our bodies and health. Computer scientists see tremendous potential in AI solutions for building on and linking existing knowledge from a vast number of scientific papers. While major investment is being made in nutritional research worldwide, many findings are hidden away in countless studies which means only limited use can be made of them.
Live at the International Food Innovation Conference
Ilias Tagkopoulos is, among other roles, Director of the USDA AI Institute for Next-Generation Food Systems (AIFS), which bridges AI and decision-making with applications in food, nutrition and health.
Food knowledge database
Tagkopoulos aims to change this situation. He and his team are using AI to develop a new food database called the 'Food Atlas'. It contains information on food, chemicals and diseases and the links between them. All this data is linked to scientific sources, allowing researchers to use it to develop new products or medical treatments. Conventional nutritional databases often don't provide these links or only contain a small proportion of all foods.
Tagkopoulos firmly believes AI-led tools are needed to collect and structure data quickly so it can be used for evidence synthesis. For example, he has used his 'Food Atlas' to establish over 230,000 links between foods and their chemical ingredients from many scientific papers. Almost half of these links were not previously documented in any other database. However, simply feeding data into the AI tool does not provide adequate results. The food industry is extremely multidimensional and highly complex. The right datasets and in-depth training of the AI tool is required to produce a model that generates reliable output.
AI – driving innovation
As part of the development of the database, AI is mainly used to improve productivity so that links can be created between the vast quantities of data available from studies. Other examples show how AI contributes to innovation in the food industry. Tagkopoulos used AI to identify a biomarker that can predict the effectiveness of a low-FODMAP diet in irritable bowel syndrome patients. While this special diet is often used to treat this condition, it does not work for all patients. To gain a proper understanding of what was going on, his lab fed results from clinical studies and microbiome data obtained from patients into a model and then used the AI tool to search for a pattern, enabling the biomarker to be identified.
The future of the food industry
AI could help us to gain a better understanding of our diet in future. By grouping foods with a similar chemical composition together, for example, we could adapt healthy diets to a particular person's health situation. Incorporating other factors, such as flavour or texture, may also enable us to find out how we can eat a healthy diet with as little sacrifice as possible. However, the situation involves much more than just the food itself in Tagkopoulos' view. A sustainable, healthy diet begins with farming: what effect do the seeds, cultivation methods and type of processing have on the health benefits later on of the chemicals found in our food? And how can the food industry's entire value chain be structured in an eco-friendly way? The researcher firmly believes that AI tools – in combination with evidence synthesis – will enable us to make major progress in these areas.
Learn more about AI as a new operating system in the food industry at the International Food Innovation Conference being held on 18 June at the GDI. Ilias Tagkopoulos and other experts from science and industry will provide strategic insights into 'Recoding Food' to explain how AI can be harnessed as a lever for resilience, growth and new value creation approaches.