AI-POWERED DIETARY RECOMMENDATION SYSTEMS FOR OPTIMIZED GLYCEMIC CONTROL IN TYPE 2 DIABETES

Abstract

Effective dietary management is essential for maintaining glycemic stability in individuals with Type 2 Diabetes, yet traditional nutritional guidelines often overlook the substantial inter-individual variability in postprandial glucose responses. This study presents an AI-powered dietary recommendation system that integrates continuous glucose monitoring data, nutritional composition features, and personalized metabolic patterns to predict glycemic excursions and generate optimized meal recommendations. Using a hybrid deep-learning architecture enhanced with temporal modeling and attention-based interpretability, the system accurately modeled nonlinear nutrient–glucose interactions and produced individualized predictions that outperformed conventional glycemic estimation approaches. A multi-objective optimization engine further refined dietary recommendations by balancing predicted glycemic impact with nutritional adequacy and patient-specific dietary constraints. Experimental results demonstrated strong predictive performance, stable confidence across diverse meal categories, and clinically meaningful nutrient–glycemic associations. Expert dietitians validated the interpretability and suitability of the recommendations, ensuring alignment with established clinical nutrition standards. While limitations include variability in CGM accuracy and incomplete meal-logging adherence, the findings provide compelling evidence that AI-driven dietary systems can significantly improve glycemic control strategies. This research highlights the promise of precision nutrition technologies and establishes a scalable, clinically relevant framework for integrating AI-based dietary recommendations into diabetes-care ecosystems.