Artificial Intelligence and Machine Learning for Advancing Dietary Assessment in National Food Intake Surveys

Background National food intake surveys are indispensable for understanding population dietary patterns, estimating nutrient intakes, identifying at-risk subgroups, and informing evidence-based nutrition policy and public health practice. These surveys support critical functions including nutrition surveillance, dietary exposure and risk assessment, food-based nutrition guideline development, and program evaluation, thereby contributing to population health surveillance and evidence-informed public health practice. However, traditional dietary assessment methods—such as dietary records, interviewer-administered 24-hour recalls, food frequency questionnaires, and weighed food records—are labor-intensive, costly, and impose substantial respondent burden. Weighed food records are generally regarded as one of the most accurate self-report dietary assessment methods; however, they require participants to weigh all foods, which can reduce compliance, limit scalability, and introduce reactivity bias. Furthermore, significant differences persist between high-income and low- and middle-income countries in survey infrastructure, frequency, regularity, and methodological capacity, with cost being the most frequently cited constraint. Emerging machine learning and artificial intelligence tools offer potential breakthroughs through image processing, natural language processing, and predictive modeling. However, many of these approaches remain insufficiently validated, standardized, and evaluated for routine implementation across diverse populations and national dietary surveillance systems.Goal The goal of this Research Topic is to advance dietary assessment methodology for national individual food intake surveys (food consumption surveys or dietary surveys), with a primary focus on evaluating machine learning and artificial intelligence (AI) applications against established dietary assessment methods and conventional survey workflows, and their potential contribution to more effective population health surveillance and public health practice. Emerging technologies—including image-based food recognition, automated portion size estimation, and large language models for dietary interview chatbots, automated food coding, extraction of structured dietary information from free-text recipes and food records, and automated mapping of foods to nutrient composition databases—offer opportunities to reduce respondent and researcher burden, lower costs, and improve more regular and frequent use of these surveys. However, challenges remain, including accurate identification of mixed dishes and culturally diverse foods, ensuring reproducibility and algorithmic transparency, addressing privacy concerns, and promoting equity across populations with varying digital literacy and access. Beyond AI-related applications, this collection also welcomes methodological contributions addressing persistent challenges in food intake/dietary surveys and nutritional epidemiology, including sampling design, interviewer training and standardization, measurement error identification and correction, bias identification, validation frameworks, harmonization across populations, and strategies to improve participation and representativeness, particularly where these contributions and innovations may inform future digital and automated approaches. By consolidating current evidence and identifying research gaps, this collection aims to support the development of validated, scalable, and cost-effective dietary surveillance systems suitable for both high-income and low- and middle-income settings.Scope We welcome quantitative, qualitative, and mixed-methods submissions addressing: (1) validation studies comparing AI- and machine learning-based dietary intake assessment tools against conventional methods and/or biomarkers; (2) pilot and feasibility studies of AI-enabled dietary intake assessment technologies, including but not limited to image recognition, natural language processing, large language models, automated food coding, and smartphone-based dietary assessment; (3) gamification and interactive approaches to improve participant engagement and reduce respondent burden; (4) hybrid implementation strategies integrating digital and traditional methods; (5) cost-effectiveness, scalability, population-level applicability, and implementation science-based evaluations of novel dietary intake assessment approaches; (6) cross-cultural adaptation and validation across diverse food cultures and settings; (7) methodological advances in measurement error correction, diet-disease association evaluation, calibration, usual intake modeling, validation frameworks, and harmonization across populations; and (8) innovations in sampling design, interviewer training, administering, survey participation, representativeness, and equity across socioeconomic and demographic groups, especially, where these innovations can strengthen population health surveillance and inform public health policy and practice. We invite original research articles, methods papers, reviews, research briefs, and perspectives from both high-income and low- and middle-income settings. Topic editors: Samandika Saparamadu, Stefanos Tyrovolas, Gregory Maniatopoulos, Lawrence J. Cheskin Source: https://www.frontiersin.org/research-topics/83757/artificial-intelligence-and-machine-learning-for-advancing-dietary-assessment-in-national-food-intake-surveys