Artificial intelligence in child nutrition and eating behavior: from prediction to gastronomic mediation

Abstract

Artificial intelligence (AI) is entering child nutrition through dietary assessment, malnutrition forecasting, clinical decision support, meal recommendation, conversational interventions, and food-environment monitoring. The consequences of these applications converge in everyday eating. This Mini Review synthesizes evidence on how AI measures nutritional states and mediates food choice, communication, sensory acceptance, family practice, and digital exposure. The evidence supports two connected functions. As nutritional intelligence, AI converts clinical, dietary, and environmental data into assessments or predictions. As gastronomic mediation, it participates in decisions about what foods are noticed, recommended, prepared, discussed, and accepted. Direct pediatric validation is strongest for bounded assessment and forecasting tasks. Child meal-planning studies and generative-AI evaluations based on standardized adolescent profiles reveal a gap between nutrient optimization, culinary coherence, and nutritional safety. A large adolescent chatbot trial combined scalable delivery with null intention-to-treat effects on diet quality and BMI trajectory. Co-design and behavioral studies further identify familiarity, texture, participation, and caregiver involvement as central design variables. We propose five iterative translational gates: technical validity, nutritional validity, behavioral acceptability, contextual legitimacy, and real-world effectiveness and implementation. Future research should combine age-specific nutritional constraints with sensory and cultural knowledge, evaluate performance across food cultures, preserve professional and caregiver oversight, and test outcomes in homes, schools, clinics, and digital food environments. AI can advance child-focused gastronomy by translating computational outputs into safe, culturally meaningful, and developmentally appropriate eating practices.