This project is an evolution in the Tech4Diet research line (www.tech4d.ua.es). It is focused on providing a integrated and personalized nutritional system. To do so, technology and nutrition has come together to create a multimodal and multidomain artificial «brain» that has information about patients that were part of previous projects, general information about nutrition and also psychological knowledge that frames the nutritional treatments.
The idea
The schematic that you can see next shows the main parts of the project. We divide it into four parts.
The main brain of the system
This part is a multimodal and multidomain space that includes information about patients that where part in previous projects. From those patients we have their biomedical parameters, their 3D avatar, and the psychological evaluation for each intervention. With that we have a large set of information about different cases of body evolution in nutritional procedures. Moreover, in this «brain» we include knowledge about what is known in nutrition scientifically, like indices, evolution of fat distribution consequences, markers of different pathologies, etc. Finally, along with the psychological information of the patients we add to this space what psychologists know about the adherence to treatments, how a certain diet affects the patient, … With all that information, this main brain will be a data consumer and provider to the rest of the system. Consuming information of new patients that are monitored, and providing information to suggestions, personalized treatments, possible new multidimensional indices for experts, etc.
Continuous monitoring
To monitor new patients it is proposed here to have a system that scans the body, acquire daily activity data, and gets the food information of each meal. Scanning the 3D model of the body has been done in previous projects using a cabin setup with multiple 3D (RGBD) cameras located around the individual. However, this cannot be applied at home, so we propose an artificial intelligence based method that allows to infer a 3D avatar from one of few colour images from a camera (smartphone or webcam). When the patient is known for the system, extra information can be taken from the main brain like previous avatars, fat distribution, etc. so the system can infer the avatar more precisely.
The food intake monitoring will be done (semi) automatically. The patient will take a picture of the meal and the application will detect the elements and estimate the quantity so an average food nutrients estimation will be done. The individual will check to
Personalized treatment
To provide with useful information to the patient in a personalized way, this module is meant to develop a realistic visualization system for mobile devices and Virtual Reality glasses. With it, the patient will be able to see their evolution in a serie of avatars with the health parameters associated to each intervention. Moreover, this module will provide recommendations for healthy habits using the information of the patient, their evolution and all the knowledge from the main brain. Alerts will also be given by the system in case dangerous behaviours or tendencies in the evolution of the patient are detected.
Funding
The project PID2023-149562OB-I00 funded by MICIU/AEI /10.13039/501100011033 and FEDER, UE
