Department of Computer Engineering

This study presents an AI-driven clinical decision support system (CDSS) that integrates advanced Large Language Models (LLMs) and federated learning within a privacy-preserving framework. The system addresses the significant challenge of processing unstructured medical data, leveraging LLMs to enhance the accuracy and relevance of diagnostic insights. Federated learning facilitates robust model training across decentralized networks while maintaining strict patient data privacy and adhering to GDPR and FAIR Data Principles. The architecture is designed to be highly scalable and adaptable, making it suitable for diverse clinical settings, ranging from large hospital networks to smaller healthcare providers. Through a comprehensive evaluation methodology, the system demonstrates the potential to significantly improve clinical decision-making by providing timely, accurate, and contextually relevant recommendations while ensuring data security and compliance with legal standards. The anticipated results suggest that the CDSS can streamline diagnostic processes, reduce errors, and support more personalized patient care. The proposed system represents a significant advancement in AI-driven healthcare solutions, aiming to enhance clinical outcomes and operational efficiency. Future research will involve real-world implementation and further development of the system's explainability and user interface to maximize its utility in clinical environments.