The integration of Deep Learning (DL) techniques with the Internet of Things (IoT) has emerged as a transformative paradigm in the advancement of smart healthcare systems. Numerous recent studies have investigated the convergence of these technologies, demonstrating their potential in improving healthcare delivery, patient monitoring, and clinical decision-making. The ongoing evolution of Industry 5.0 in parallel with the deployment of 5G communication networks has further facilitated the development of intelligent, cost-effective, and highly responsive sensors. These innovations enable continuous and real-time monitoring of patients’ health conditions, a capability that was not feasible within the constraints of traditional healthcare models. Smart health monitoring systems have thus introduced significant improvements in terms of speed, affordability, reliability, and accessibility of medical services, particularly in remote or underserved regions. Moreover, the application of Deep Learning and Machine Learning algorithms in health data analysis has played a pivotal role in achieving preventive healthcare, reducing mortality risks, and enabling personalized treatment strategies. Such methods have also enhanced the early detection of chronic diseases, which previously posed considerable diagnostic challenges. To further optimize scalability and cost-efficiency, cloud computing and distributed storage solutions have been incorporated, ensuring secure and real-time data availability. This review therefore provides a comprehensive perspective on smart healthcare innovations, emphasizing the role of intelligent systems, recent advancements, and persisting challenges in the domain of digital health monitoring.
The growing sophistication of cyber threats exposes the limits of signature-based detection in Security Information and Event Management (SIEM) systems. User and Entity Behavior Analytics (UEBA) advances SIEM by enabling behavior-based anomaly detection, yet legacy approaches struggle with high false positives and poor adaptability to evolving threats. This research proposes an AI-driven UEBA framework that combines deep learning for modeling user behavior with graph-based tools to map system relationships, enhancing anomaly detection in enterprise environments. Using datasets such as CERT Insider Threat, UNSW-NB15, and TON_IoT, we simulate diverse behaviors and evaluate performance. Our Transformer-GNN ensemble achieved an F1-score of 0.90, reduced false positives by 40%, and cut incident triage time by 78% compared to rule-based SIEM. To support real-world use, we provide an open-source pipeline integrating with SIEM platforms via Kafka, Elastic search, and a modular ML inference layer. This work bridges AI research and deployable cybersecurity practice, advancing the development of adaptive, intelligent, and robust UEBA systems.