DESIGN AND IMPLEMENTATION OF A DIETARY SYSTEM FOR DIABETIC PATIENT

ABSTRACT This project designed and implemented a personalized dietary system for diabetic patients, leveraging machine learning to provide culturally adapted meal recommendations tailored to Nigerian users. The growing prevalence of diabetes in Nigeria, coupled with limited access to nutritionists and the cultural irrelevance of existing dietary apps, necessitated an intelligent solution that bridges nutritional science with local dietary practices. The system employed a hybrid recommendation algorithm combining collaborative filtering and content-based approaches to analyze user preferences, health conditions, and ingredient restrictions. A user-friendly web interface was developed using React.js (frontend) and Node.js (backend), integrated with a MySQL database of Nigerian recipes and nutritional data. Key functionalities included ingredientbased filtering, dynamic adaptation to user feedback, and offline accessibility for low-resource settings. Testing demonstrated 85% accuracy in aligning recommendations with users’ health goals and dietary constraints, while user surveys reported 90% satisfaction with cultural relevance. The system addressed critical gaps in existing platforms by prioritizing local staples (e.g., garri, plantains) and Nigerian meal structures, proving particularly effective for diabetic patients requiring carbohydrate-controlled diets. Challenges included sparse nutritional datasets for Nigerian foods and computational costs of real-time personalization. This work contributes to digital health innovation in low-resource contexts by demonstrating how machine learning can be adapted to non-Western dietary ecosystems. Future enhancements could integrate clinical data validation and multilingual support for broader accessibility.