Artificial Intelligence-Assisted Nutritional Counseling for Weight Management in Adult Obesity: A Systematic Review of Randomized Controlled Trials and Primary Studies

Abstract

Introduction: The global obesity pandemic affects over 2.11 billion adults, with projections exceeding 3.80 billion by 2050. Conventional nutritional counseling modalities face critical limitations in scalability, individualization, and long-term adherence. Artificial intelligence (AI)-assisted nutritional counseling, employing machine learning, deep learning, natural language processing, and reinforcement learning algorithms, represents a transformative paradigm enabling personalized, real-time dietary interventions at population scale. Methods: This systematic review was conducted following PRISMA 2020 guidelines. Eligible studies comprised randomized controlled trials and primary prospective studies evaluating AI-assisted nutritional interventions in adults (≥18 years) with overweight (BMI ≥25 kg/m²) or obesity (BMI ≥30 kg/m²). Fifteen pre-specified outcome domains were evaluated. Risk of bias was assessed using Cochrane RoB 2.0 and ROBINS-I tools. Results: Eighteen studies (n=14,732 participants) met eligibility criteria. AI-assisted interventions demonstrated statistically significant reductions in body weight (MD −1.60 kg to −12.3%; 15/18 studies, p<0.01), BMI (MD −0.59 to −1.26 kg/m²; 12/14 studies, p<0.05), HbA1c (MD −0.28% to −2.9%; 9/11 studies), and fasting plasma glucose (7/8 studies). Significant improvements were observed across cardiometabolic, anthropometric, dietary quality, physical activity, health-related quality of life, and patient engagement domains. AI-led interventions demonstrated non-inferiority to human coaching and superiority to standard care. Discussion: AI-assisted nutritional counseling demonstrates robust, clinically meaningful efficacy across multiple outcome domains. Precision nutrition algorithms integrating postprandial glycemic response prediction and gut microbiome profiling achieved the most comprehensive cardiometabolic optimization. AI-integrated mHealth applications demonstrated superior population-level scalability. Methodological limitations include heterogeneity across AI modalities, short intervention durations, and underrepresentation of low- and middle-income country populations. Conclusion: AI-assisted nutritional counseling is efficacious, safe, scalable, and non-inferior to human coaching for adult obesity management across ≥10 clinically relevant outcome domains. Integration into multidisciplinary obesity management clinical pathways is recommended, supported by long-term, equity-focused randomized controlled trials with standardized outcome reporting frameworks.