Comprehensive Review on Artificial Intelligence-Based Smart Healthcare Monitoring Using IoT and HCI

Abstract

The convergence of the Internet of Things (IoT), artificial intelligence (AI), and human–computer interaction (HCI) is reshaping healthcare monitoring from episodic clinical observation toward continuous, context-aware, and increasingly personalized health management. IoT-enabled medical sensors, wearable devices, ambient intelligence systems, mobile platforms, and connected clinical equipment generate heterogeneous physiological, behavioural, environmental, and contextual data at unprecedented temporal resolution. AI techniques transform these data into predictions, classifications, anomaly detection, risk scores, and decision-support information, while HCI provides the mechanisms through which patients, caregivers, and clinicians perceive, interpret, validate, and act upon algorithmic outputs. Recent reviews indicate that AI-enabled remote monitoring has expanded from conventional cloud-based architectures toward edge intelligence, federated learning, explainable AI, and human-centred Healthcare 5.0 architectures [1]–[5]. This review critically examines the integration of AI, IoT, and HCI in smart healthcare monitoring. A layered taxonomy is developed covering sensing and acquisition, communication, edge/cloud intelligence, clinical analytics, interaction, and governance. Machine-learning, deep-learning, multimodal fusion, time-series modeling, explainable AI, reinforcement learning, and federated-learning methodologies are examined with respect to their suitability for continuous monitoring. Particular attention is given to latency, energy consumption, model generalization, data heterogeneity, interpretability, interoperability, privacy, cybersecurity, usability, and clinical validation. The review also compares centralized, edge, cloud, and federated architectures and examines their implications for real-time healthcare applications. Representative applications involving cardiovascular monitoring, diabetes management, respiratory disease, neurological disorders, elderly care, rehabilitation, mental-health monitoring, and emergency detection are discussed. A reference case study is presented for an AI-assisted remote patient-monitoring system integrating wearable sensors, edge analytics, a clinical dashboard, and human-centred alert mechanisms. Finally, the review identifies unresolved challenges involving dataset bias, sensor reliability, model drift, human over-reliance, alarm fatigue, interoperability, regulatory compliance, and equitable access. Future research directions include multimodal foundation models, adaptive edge intelligence, digital twins, privacy-preserving learning, standardized evaluation protocols, explainable multimodal AI, and human-AI collaborative decision-making.