Edge-to-Cloud Data Engineering Pipelines for Real-Time Healthcare Predictive Analytics

Authors: Sai Kiran Yadav Battula

DOI: https://doi.org/10.37082/IJIRMPS.v13.i6.232863

Short DOI: https://doi.org/hbf8zg

Country: United States

Full-text Research PDF File:   View   |   Download

Abstract: The rapid proliferation of Internet of Medical Things (IoMT) devices and electronic health records (EHRs) has created continuous, high-velocity patient data streams with the potential to transform reactive care into proactive, data-driven intervention [1]. Realizing this vision requires edge-to-cloud data engineering pipelines that can securely ingest, standardize, and analyze heterogeneous, privacy-sensitive data under strict latency and regulatory constraints [3], [7]. This paper presents an interoperable Dual-Loop Edge-to-Cloud Data Engineering Pipeline that reconciles real-time actuation with longitudinal learning [3]. A Fast Path performs ultra-low-latency edge inference and triage, while a Slow Path supports cloud-scale training and retrospective analytics [3], [7]. A Cross-Domain Data Fabric coordinates both loops by enforcing semantic interoperability through HL7 FHIR-aligned modeling and active metadata-driven routing [2], [8], [9]. To address waveform interoperability bottlenecks, we introduce a recursive FHIR SampledData flattening and SIMD-accelerated columnarization technique that converts nested physiological resources into analytics-ready Parquet/Delta formats at sustained real-time throughput [8], [9]. Privacy is ensured via a federated learning workflow augmented with Differential Privacy (ε≈8) and Secure Multi-Party Computation (SMPC) aggregation, yielding formal privacy guarantees with negligible clinical utility loss [10], [11]. A cardiac monitoring case study demonstrates reliable sub-300 ms end-to-end alerting, ~98% upstream bandwidth reduction, and clinically actionable arrhythmic event prediction, validating the architecture’s suitability for life-critical decision support [4], [5], [7].

Keywords: Edge computing; Internet of Medical Things (IoMT); real-time predictive analytics; HL7 FHIR; streaming analytics; cardiac monitoring; federated learning; differential privacy; secure aggregation; data fabric; zero trust security.

Paper Id: 232863

Published On: 2025-11-28

Published In: Volume 13, Issue 6, November-December 2025

Share this