Towards collaborative intelligent IoT eHealth: From device to fog, and cloud

The relationship between technology and healthcare due to the rise of intelligent Internet of Things (IoT), Artificial Intelligence (AI), and the rapid public embracement of medical-grade wearables has been dramatically transformed in the past few years. AI-powered IoT enabled disruptive changes and unique opportunities to the healthcare industry through personalized services, tailored content, improved availability and accessibility, and cost-effective delivery. Despite these exciting advancements in the transition from clinic-centric to patient-centric healthcare, many challenges still need to be tackled. The key to successfully unlock and enable this horizon shift is adopting hierarchical and collaborative architectures to provide a high level of quality in key attributes such as latency, availability, and real-time analytics. In this paper, we propose a holistic AI-driven IoT eHealth architecture based on the concept of Collaborative Machine Learning approach in which the intelligence is distributed across Device layer, Edge/Fog layer, and Cloud layer. This solution enables healthcare professionals to continuously monitor health-related data of subjects anywhere at any time and provide real-time actionable insights which ultimately improves the decision-making power. The feasibility of such architecture is investigated using a comprehensive ECG-based arrhythmia detection case study. This illustrative example discusses and addresses all important aspects of the proposed architecture from design implications such as corresponding overheads, energy consumption, latency, and performance, to mapping and deploying advanced machine learning techniques (e.g., Convolutional Neural Network) to such architecture.