Case Study

Enhancing clinical nutrition monitoring with AI, Validithi Project

The Validithi Project aimed to evaluate the accuracy and feasibility of using AI-powered food recognition to automate dietary intake analysis in clinical settings.

Background & Objectives

Inadequate nutrition is a leading contributor to noncommunicable diseases (NCDs) such as cardiovascular disease, diabetes, and cancer, significantly driving global healthcare costs. Traditional dietary assessment methods—such as 24-hour recalls, food diaries, FFQs, and dietary interviews—are manual, time-consuming, and error-prone. These methods are heavily dependent on memory and estimation, often leading to inconsistencies in portion size and food content reporting, with error margins of up to 35%.

The Validithi Project aimed to evaluate the accuracy and feasibility of using AI-powered food recognition to automate dietary intake analysis in clinical settings. The project sought to:

Reduce time and costs associated with manual dietary assessment
Improve the precision of nutritional tracking
Enhance patient motivation and engagement through real-time feedback
Validate LogMeal AI food logging accuracy against biological markers (urinary recall)

Initial Assessment

Conducted in collaboration with hospitals and academic partners in the Netherlands
Participants were assessed using traditional 24-hour food recall protocols
Nutrient intake from these recalls was cross-validated with urinary biomarkers (objective biological data)
Participants also logged meals using LogMeal’s food recognition algorithms, with AI-driven identification of foods, ingredients, and portion sizes

Digital Monitoring with LogMeal

Automated photo-based food logging: Participants captured images of meals in real time
AI-powered analysis: Nutritional breakdown of calories, macronutrients, micronutrients, and ingredients using LogMeal’s 37-nutrient model
Automatic food diary generation, eliminating the need for manual entries
Comparison with urine-based nutritional analysis to validate precision of the AI system
Clinician dashboard access for real-time intake reports and nutrient analysis

Group and Personalized Interventions

Healthcare providers accessed automated dietary logs for individual consultations
Real-time data enabled faster feedback and dietary adjustments
Data used to enhance motivational strategies, showing patients quantified nutrition insights tied to biomarkers
Improved efficiency of care through reduced workload for dietitians and minimized consultation time

With all this in mind, in this project we compared the use of LogMeal’s automated food recognition log with the nutritional values extracted by a 24-hour recall urine test. Here you can see the comparison of the results of one of the European projects we did with hospitals and universities in the Netherlands.

Selected Food-Related Papers:

paper_logmeal_1.pdf , paper_logmeal_2.pdf , paper_logmeal_3.pdf

Listing of Dr.a Petia Radeva’s (Full Professor UB) repository of AI Articles, Computer Vision, by Dr.a Petia Radeva
https://dblp.org/pid/r/PetiaRadeva.html

Papers List AI, Computer Vision, from Dr. Marc Bolaños PhD in Computer Vision “Deep Multimodal Learning for Egocentric Storytelling and Food Analysis”
https://dblp.org/pid/147/3344.html

Dr. Petia Radeva, professor and head of the Computer Vision Department at U.B. (University of Barcelona).
http://www.ub.edu/cvub/petiaradeva/?p=36

Results and value of Logmeal in the program

Reduction in consultation time: Nutrient analysis performed automatically, saving 15–30 minutes per patient
Higher data accuracy: Significant reduction in self-reporting bias and portion misestimations
Better treatment adherence: Patients responded positively to data-driven feedback and visual dietary reports
Clinical-grade reliability: AI intake analysis correlated strongly with 24-hour urine biomarkers
Image-based dietary analysis with instant recognition of foods, portions, and ingredients

LogMeal Technology Value in the Program

39-nutrient breakdown enabling granular tracking of micro- and macronutrients
Seamless integration into clinical workflows, enabling professional oversight without extra workload
Automated food diaries to enhance accuracy and reduce human error
Enhanced patient empowerment through tangible, visual health metrics

Conclusion

The Validithi Project demonstrated that AI-powered food recognition tools like LogMeal can effectively complement or even replace traditional dietary assessments in clinical practice. The use of objective image-based logging aligned closely with biological nutrient indicators, while significantly reducing the burden on healthcare professionals and increasing engagement among patients.

Reduction of up to 35% error rate from self-reported dietary surveys
Saves 15–30 minutes per consultation by replacing manual intake assessments
Increases accuracy and consistency in monitoring macro- & micronutrient intake
Enhances patient motivation and treatment adherence through visual & data-driven feedback
LogMeal’s automated food recognition offers a more efficient, scalable, and objective alternative to traditional intake tracking, validated through European hospital and university collaborations.

partners and collaborators

HEALTH & NUTRITION REFERRALS

Stephan Bakker
Professor of Internal Medicine at University Medical Center Groningen (UMCG)