Related Experiment Video
Updated: Jan 8, 2026
06:21
Concept Development and Use of an Automated Food Intake and Eating Behavior Assessment Method
Published on: February 19, 2021
6.2K
Artificial intelligence applications for assessing ultra-processed food consumption: a scoping review
Jessica L Campbell1, Grant Schofield1, Hannah R Tiedt2
1Human Potential Centre, Faculty of Health and Environmental Sciences, https://ror.org/01zvqw119Auckland University of Technology, Auckland1142, New Zealand.
The British Journal of Nutrition
|December 21, 2025
Summary
Artificial intelligence (AI) can classify ultra-processed foods (UPFs) using various techniques. This review maps AI applications in UPF classification, showing potential for improved dietary monitoring and public health strategies.
Area of Science:
- Nutrition Science
- Computer Science
- Public Health
Background:
- Ultra-processed foods (UPFs) are linked to adverse health outcomes.
- Identifying and monitoring UPF consumption is crucial for public health.
- The application of artificial intelligence (AI) in classifying UPFs is underexplored.
Purpose of the Study:
- To conduct a scoping review on the use of AI in classifying ultra-processed foods (UPFs).
- To map AI techniques, input data, classification frameworks, accuracy, and applications for UPF assessment.
- To identify the potential of AI in improving dietary monitoring.
Main Methods:
- Systematic search of scientific literature (PubMed, Scopus, Medline, CINAHL) from 2015-2025.
- Inclusion of peer-reviewed studies applying AI to assess or classify UPFs using recognized frameworks.
- Independent screening and data extraction by two reviewers, focusing on AI methods, input data, frameworks, and outcomes.
Main Results:
- Eight studies met inclusion criteria, utilizing diverse AI techniques like random forest classifiers and large language models.
- Four studies explored practical applications, including assessing consumption behaviors and developing healthier substitution tools.
- Reported predictive accuracies (F1 scores) ranged from 0.86 to 0.98, with good alignment between AI-derived clusters and NOVA categories.
Conclusions:
- AI tools demonstrate significant potential for enhancing the monitoring of ultra-processed food consumption.
- Further development and validation of real-time AI methods are needed to support public health initiatives.
- AI offers a promising avenue for objective and scalable UPF classification.
