The integration of artificial intelligence (AI) and the Internet of Things (IoT) in healthcare has revolutionized real-time patient monitoring through wearable devices. Leveraging AI-driven analytics, these systems can detect anomalies in vital signs and enable early disease prediction with unprecedented accuracy. This paper proposes an innovative AI-powered IoT wearable framework that incorporates state-of-the-art deep learning models, such as convolutional neural networks (CNNs), long short-term memory (LSTM) networks, and transformers, for real-time data analysis. By integrating edge computing, the system performs on-device processing to minimize latency and reduce computational load. To address privacy concerns, federated learning (FL) is employed, enabling decentralized training of AI models while keeping sensitive patient data secure. Validated on physiological sensor datasets, the proposed architecture demonstrates significant improvements in diagnostic precision, energy efficiency, and compliance with healthcare data privacy regulations. The outcomes underscore the transformative potential of AI-enhanced wearable technologies for continuous health monitoring, timely anomaly detection, and robust, privacy-preserving patient data management in remote healthcare environments.