Asynchronous proximal federated aggregation for heterogeneous healthcare networks - Takeaways - MDSpire

Asynchronous Proximal Federated Aggregation Framework for Diverse Healthcare Networks

  • By

  • Manakkattu Sreelakshmi

  • Radhakrishnan Delhibabu

  • July 17, 2026

Share

  • 1

    The Asynchronous Proximal Federated Aggregation (APFA) framework addresses computational asymmetry and statistical heterogeneity in Federated Learning.

  • 2

    APFA allows continuous model updates from edge devices, integrating a local proximal regularizer and a server-side staleness dampening penalty.

  • 3

    In evaluations on CheXpert and MIMIC-IV datasets, APFA achieved an 80% diagnostic viability threshold in 4.1 simulated hours.

  • 4

    APFA demonstrated a 71% reduction in total wait time compared to standard synchronous baselines like FedProx.

  • 5

    The framework effectively mitigates weight divergence, showcasing the potential of asynchronous machine learning in clinical diagnostics.

Original Source(s)

Asynchronous proximal federated aggregation for heterogeneous healthcare networks

  • by Manakkattu Sreelakshmi, Radhakrishnan Delhibabu

  • July 17, 2026

Related Content

Biological effects of 18F-FDG administration and CT dose during PET/CT: chromosomal aberrations in a two-center prospective observational study

Non-linear association between thyroid scintigraphy-derived thyroid weight and I-131 treatment efficacy in graves’ disease: a multicenter restricted cubic spline and threshold analysis

Risk-adapted trajectory selection for ultrasound-guided thyroid fine-needle aspiration biopsy: a multicenter study of in-plane and out-of-plane approaches