Artificial Intelligence in Dietary Management and Nutritional Interventions: A Narrative Review of Current Evidence, Applications, and Translational Challenges

This narrative review examines the applications of artificial intelligence (AI) in nutrition and dietetics, focusing on AI-assisted dietary assessment, precision nutrition, generative AI for dietary counselling, and the ethical, regulatory, and implementation challenges affecting clinical translation. A critical narrative synthesis of evidence on AI applications in nutrition and dietetics was undertaken, covering convolutional neural network (CNN)-based food image recognition and portion estimation, machine learning (ML) models for predicting postprandial glycaemic response (PPGR), and large language model (LLM)-based chatbots for dietary counselling, meal planning, and behaviour change. Evidence on clinical outcomes, model performance, validation, and implementation barriers was considered. AI-assisted dietary assessment may improve accessibility and achieve performance comparable to human assessment for portion and energy estimation under controlled conditions; however, limited external validation and under-representation of diverse foods constrain generalisability. ML models integrating anthropometric, clinical, dietary, and gut microbiome data can improve individual PPGR prediction, with a limited number of randomised trials reporting beneficial glycaemic or metabolic outcomes from ML-guided personalised diets. LLM-based chatbots can provide broadly satisfactory dietary guidance and support conversational engagement, but inaccuracies, micronutrient omissions, and lower empathic quality than human dietitians remain concerns. Evidence from rigorous randomised controlled trials with clinically meaningful, long-term outcomes is limited. Additional barriers include algorithmic bias, data privacy and security risks, limited explainability, and evolving regulatory frameworks. AI has considerable potential to support dietary assessment, personalised nutrition, and nutrition counselling, but current evidence does not justify replacing registered dietitians and nutrition professionals. Standardised validation, rigorous clinical trials, improved model transparency, and equity-focused governance are essential to enable safe, effective, and responsible integration of AI into routine nutrition practice.