MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202641097680 A) filed by J. J College Of Engineering And Technology on August 12, 2026, for Adaptive Federated Deep Learning Framework For Privacy-Preserving Cancer Diagnosis Across Distributed Hospital Networks.
Inventors include Dr S Murugesan; Dr I Shahanaz Begum; Dr. K. Saravana Kumar; Jaison Vimalraj T; and S Lakshmi Narasimhan.
The application for the patent was published on August 21, 2026, under issue no. 34/2026.
Abstract: The present invention discloses an adaptive federated deep learning framework for privacy-preserving cancer diagnosis across distributed hospital networks. The framework comprises a plurality of hospital computing nodes and a federated coordination node configured to collaboratively train a cancer diagnosis model without transferring raw patient datasets from respective hospitals. An adaptive client management module selectively determines hospital participation based on data quality, computational capability, communication condition, model performance and distribution characteristics. Local model updates are privacy-protected before transmission and are adaptively aggregated using contribution weights determined from local validation performance, data quality, distribution divergence and update reliability. A distribution monitoring module detects changes in local data characteristics and dynamically modifies training, aggregation and personalization parameters. Hospital- specific model parameters are retained to adapt a shared global model to heterogeneous clinical environments. The framework thereby provides distributed cancer model training with reduced raw-data exposure, adaptive handling of heterogeneous hospital datasets, communication-efficient participation and locally personalized diagnostic inference.
Disclaimer: Curated by HT Syndication.
