Big Data Analytics for Remote Patient Monitoring in Cyber-Physical Healthcare Systems
TL;DR
A six-layer Big Data analytics framework for RPM in cyber-physical healthcare settings is proposed, integrating IoMT wearable sensing, edge computing preprocessing, Apache Kafka stream ingestion, Apache Spark distributed processing, and a hierarchical Long Short-Term Memory deep learning model for real-time anomaly detection.
Abstract
The convergence of Big Data analytics with Cyber-Physical Systems (CPS) presents a transformative opportunity for Remote Patient Monitoring (RPM) in modern healthcare. Despite advances in the Internet of Medical Things (IoMT) and distributed computing, the existing RPM frameworks lack integrated support for real-time multi-modal physiological analysis at scale. This paper proposes a six-layer Big Data analytics framework for RPM in cyber-physical healthcare settings, integrating IoMT wearable sensing, edge computing preprocessing, Apache Kafka stream ingestion, Apache Spark distributed processing, and a hierarchical Long Short-Term Memory (LSTM) deep learning model for real-time anomaly detection. The key contributions of the current paper include: (i) a scalable six-layer CPS architecture for continuous physiological monitoring; (ii) a distributed Big Data pipeline achieving sub-2-second end-to-end alert latency, and (iii) an LSTM-based classifier supporting eight physiological condition classes with high accuracy. Experimental evaluation on a real-participant dataset demonstrates competitive performance, with low latency and strong clinician usability scores, indicating promising readiness for broader clinical evaluation and deployment.