FedFusionNet++ is a privacy-preserving Federated learning framework for Oral Cancer Detection. It combines SCAFFOLD, FedProx, and AFA with ε=3.0 differential privacy to achieve 89% accuracy on Non-IID datasets—a 23.6% improvement. The hybrid architecture integrates dual-branch CNN and XGBoost ensemble for robust clinical predictions.
transformers survival-analysis digital-pathology vlm wsi federated-learning explainable-ai swin-transformer cancer-screening kaplan-meier-curve medical-diagnosis-ai hipaa-compliant cross-vit
-
Updated
May 17, 2026 - Python