EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Acuity-Aware Federated Learning Over 6G Networks for Real-Time Patient Risk Prediction in Mobile EMS Fleets

Mir Hassan, Giedrė Sabaliauskaitė, Mabrook S. Al-Rakhami, Atif M. Alamri

IEEE Open Journal of the Communications Society · 2026 · 7 · 13924-13938

DOI: 10.1109/ojcoms.2026.3738928 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Open Journal of the Communications Society, 2026. Verfasst von Mir Hassan und 3 weiteren Autor:innen. Die Arbeit nennt 42 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Research Council of Lithuania.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Federated learning (FL) enables collaborative intelligence across distributed emergency medical services (EMS) while preserving patient privacy. However, existing FL algorithms primarily optimize communication according to client availability, update freshness, or statistical utility, assuming that all participating clients have equal operational importance. This assumption is unsuitable for mission-critical emergency response, where communication resources are limited and delayed model synchronization may disproportionately affect critically ill patients. This paper proposes an acuity-aware communication scheduling framework for FL over heterogeneous 6G-enabled EMS networks. The proposed framework jointly incorporates patient acuity, communication connectivity, and model staleness into client scheduling and model aggregation, enabling communication resources to be allocated according to clinical urgency. A lightweight and clinically interpretable acuity score is computed onboard each ambulance and used to prioritize model transmission under constrained wireless connectivity. The framework is evaluated using the PhysioNet Challenge 2019 dataset under full coverage, mixed, and stress communication conditions representing realistic EMS deployments. Experimental results show that the proposed framework improves acuity-aware communication prioritization, particularly by reducing staleness for high-acuity clients, while maintaining predictive performance comparable to conventional FL methods. Although the proposed method does not achieve the lowest fleet-average acuity-weighted staleness in every connectivity regime, it provides stronger protection against synchronization delay for clinically urgent clients under constrained communication. The results further show that application-aware communication scheduling improves communication efficiency without modifying the underlying prediction model. This work establishes a practical link between clinical decision support and communication-layer resource allocation, providing a communication-aware FL framework for future 6G-enabled mission-critical emergency response systems.

Förderung

  • Research Council of Lithuania

Zitieren

APA

Hassan, M., Sabaliauskaitė, G., Al-Rakhami, M. S., & Alamri, A. M. (2026). Acuity-Aware Federated Learning Over 6G Networks for Real-Time Patient Risk Prediction in Mobile EMS Fleets. IEEE Open Journal of the Communications Society, 7, 13924-13938. https://doi.org/10.1109/ojcoms.2026.3738928

Vancouver

Hassan M, Sabaliauskaitė G, Al-Rakhami MS, Alamri AM. Acuity-Aware Federated Learning Over 6G Networks for Real-Time Patient Risk Prediction in Mobile EMS Fleets. IEEE Open Journal of the Communications Society. 2026;7:13924-13938. doi:10.1109/ojcoms.2026.3738928

BibTeX
@article{Hassan2026,
  author = {Hassan, Mir and Sabaliauskaitė, Giedrė and Al-Rakhami, Mabrook S. and Alamri, Atif M.},
  title = {Acuity-Aware Federated Learning Over 6G Networks for Real-Time Patient Risk Prediction in Mobile EMS Fleets},
  journal = {IEEE Open Journal of the Communications Society},
  year = {2026},
  volume = {7},
  pages = {13924--13938},
  doi = {10.1109/ojcoms.2026.3738928}
}
Kostenlos starten →