Validation of an Innovative Dietary Intake Tool for Healthcare Implementation
The VALIDITHI project will prove the value of LogMeal, a tool that identifies food and the nutrients it contains from a simple smartphone photo, giving users a nutritional diary that they can share with medical professionals. LogMeal is far more accurate and versatile than other attempts to develop this technology, and its ease of use means people are more willing to adhere to the programme.
Chronic disease is responsible for the majority of healthcare costs, and half of these diseases are the result of unhealthy habits, with nutrition being a key factor. Compelling evidence shows nutrition’s ability to prevent, and even cure, disease, but many people have an inadequate food intake. Improving nutritional care via more accurate dietary intake assessment can improve peoples’ quality of life and care.
- UMCG: Entrepreneurial clinicians.
- University of Barcelona: Developers of deep learning algorithms for food image analysis.
- Centre de Visió per Computador: Engineers applying computer vision technology.
- Asc Academics: Health economics expertise to define different markets and sharpen the business model.
- Nestlé Research: Nutritional experts, software developers and business mentors.
VALIDITHI is a project to validate a food-intake monitoring tool called LogMeal, which allows users to keep automatic food diaries by simply taking smartphone photos of their food. The application recognises the food in terms of dish, category and ingredients, and the nutrient composition is calculated based on national food composition databases. The application’s algorithms achieve a very high accuracy rate of 88%.
The application will be linked to a system that lets users monitor their own nutritional intake, suggests personalised meal plans and shares the data with nutritional professionals, who can design an individualised strategy for nutritional behaviour change via continuous feedback, education and motivation.
VALIDITHI will validate LogMeal and the system’s image analysis algorithms, which are based on two promising technologies: deep learning and computer vision. The main validation study will take place in University Medical Center Groningen. The study cohort consists of renal transplant recipients and their healthy donors – a specific population characterised by high-frequency hospital visits. Proper nutrition is a key factor for successful long-term survival with a transplanted kidney. The LogMeal application will be validated against 24-hour urinary biomarkers and dietary recalls. Alongside the medical validation, health economic market analyses and business models will show the impact and business opportunities of LogMeal.
Unlike current techniques for keeping a food diary, LogMeal is easy to use and accurate, offering a system that really works, and allows medical professionals to help their patients address nutritional issues. Better nutrition allows patients to avoid many of the most common chronic diseases and stay healthy. The benefits for payers and society will be cost savings from a reduction in chronic conditions. The new technologies validated by this project can also be used in other innovations.
Activities in 2019During 2019 we've organized several dissemination activities about our goals:
|1||ICMV||16-18 of November, 2019||“Uncertainty Modeling for Improving Food Recognition"|
|2||Data Council||2nd of October, 2019||Uncertainity-Aware Food Recognition by Deep Learning|
|3||Invited talk||25th of September, 2019||Deep learning - science, technology or society solution?|
|4||Invited talk||19th of August, 2019||Uncertainity-aware Food image analysis|
|5||IBPRIA||2-3 of June, 2019||Food Recognition by Integrating Local and Flat Classifiers|
|6||Invited Seminary||15-28 of April, 2019||Uncertainity-aware CNNs. Application to food image analysis|
|7||Meeting||June, 2019||EIT Food - Quisper meeting|