ElKafrawy, PassentAlMarhbi, AraaAbu Rub, ArwaSuroor, Duaa2025-06-242025-06-242025https://repository.effatuniversity.edu.sa/handle/20.500.14131/2151Diabetes has become a significant global health concern, demanding innovative solutions for early detection and effective management. Traditional machine learning models typically rely on centralized data aggregation, raising substantial privacy issues and compliance risks. This project, Tayaqn: An AI-Driven Diabetes Prediction System Using Federated Learning Architecture, introduces a novel framework that employs Federated Learning (FL) to overcome these challenges. FL allows for secure, decentralized model training, ensuring sensitive health data remains local while enabling collaborative learning across diverse datasets. The system integrates advanced machine learning and deep learning techniques with robust privacy-preserving mechanisms, achieving high predictive accuracy while safeguarding data security. By combining scalability, regulatory adherence, and user-friendly design, this project provides a transformative solution for enhancing early diagnosis and personalized diabetes management, equipping both patients and clinicians with actionable insights for better healthcare decision-making.enCapstone