Constructing personalised nutrition plans for adolescents using knowledge graphs and dietary profiles.

INTRODUCTION: Adolescence is a crucial period for youth development. However, many teenagers consume diets high in sugar and fat but low in essential nutrients, which can impair growth and increase the risk of obesity and metabolic diseases. This study developed a knowledge graph-based system that generates personalised menus for adolescents by combining structured nutrition data with individual dietary profiles, and evaluated its effects in a school setting. METHODS: An adolescent dietary knowledge graph was built to integrate 339 dishes, 246 ingredients, 30 nutrients and 26 adolescent nutrient problems. This framework was enriched with 300 individual dietary profiles capturing physiological traits, health risks and taste preferences. Furthermore, a rule-based engine generated daily menus, aiming to meet nutrient targets. This was assessed using diet-quality indicators and a three-month randomised controlled trial involving 400 adolescents (intervention: n = 200; control: n = 200). RESULTS: Recommended menus consistently showed higher scores on the Chinese Healthy Diet Index, Dietary Diversity Score and Food Variety Score than the non-intervention meals (P < 0.0001). Following the intervention, the personalised menu group demonstrated significantly lower increases in body weight, body mass index and body fat index compared to the control group (P < 0.05). Height gain did not differ significantly between groups (P = 0.088). CONCLUSION: These results suggest that knowledge graph-based personalised recommendation systems can translate established nutritional guidelines into structured, individualised dietary guidance for adolescents. This integrated approach provides a feasible digital framework for implementing guideline-concordant meal planning in school settings and may help bridge the gap between general recommendations and the diverse needs of growing teenagers.