Artificial intelligence applications for dietary assessment in the nutrition research

Background: Multiple applications of artificial intelligence (AI) in medical sciences are growingrapidly in the recent years. AI technologies became complementary to the food science andnutrition research areas in the late ۲۰۱۰s. AI provides new opportunities for research on nutrientsand medical sensing technology. The application of AI in the nutritional epidemiology field, inparticular, dietary assessment, has been reported in several recent studies, however, any studydidn’t summarize comprehensively these findings. This systematic review aimed to provide anoverview of the main and latest applications of AI in dietary assessment research and identifygaps to address to potentialize this emerging field.Methods: This study were conducted with considering the PRISMA guidelines. The literaturesearch was conducted in PubMed, Scopus, and Google Scholar without date restriction up to February۲۰۲۳. The search strategy was expanded using a combination of MeSH terms and the followingkeywords: “artificial intelligence” AND “dietary assessment” OR “nutrient”. Moreover, amanual search of the references list of eligible studies and the Google was done to minimize therisk of missing relevant papers. All original articles written in English that evaluated the applicationof AI for dietary assessment of participants were eligible for the present review.Results: After screening the title, abstract, and full text of obtained articles by two independentauthors, finally, ۹ studies were included in the current review. The included studies were publishedfrom ۲۰۰۸ to ۲۰۲۲. The used predominant algorithms in included studies were machine learningand deep learning to estimate food portion size and estimate the calorie and macronutrient contentof a meal. Moreover, the included studies suggested the use of smartphone and image-based andweb-based dietary assessment apps in nutritional epidemiology.Conclusions: AI-based approaches including mobile apps and image recognition can improvedietary assessment by addressing random errors in self-reported measurements of dietary intakes.Further research is needed to identify and develop new AI-based approaches for dietary assessmentin nutrition research. Furthermore, well-designed studies with large sample sizes are requiredto confirm the beneficial health outcomes of AI use among different age groups of thepopulation.

کلیدواژه ها:

Artificial intelligence ، Dietary assessment ، Nutrition ، Nutrient