Federated Few-Shot Learning for Epileptic Seizure Detection Under Privacy Constraints

2025-12-09

AI Summary

Addressing the challenge of seizure detection under strict medical privacy constraints—where data across institutions exhibit non-IID distributions and each patient has only five labeled EEG segments—we propose the first Federated Few-Shot Learning (FFSL) framework. FFSL comprises two stages: (1) federated learning to collaboratively optimize a biosignal-specific Transformer (BIOT) for robust representation learning, and (2) client-side, patient-level personalization using minimal labeled samples. This paradigm enables cross-institutional knowledge sharing while preserving strict data locality. Experiments demonstrate that FFSL significantly outperforms pure federated fine-tuning—achieving balanced accuracy (0.77), Cohen’s kappa (0.62), and weighted F1-score (0.73) versus 0.43 across all metrics—and closely approaches centralized training performance (0.52). FFSL thus establishes a novel, privacy-preserving paradigm for few-shot diagnosis of neurological disorders in real-world clinical settings.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Computer Vision: Bias, Fairness & Privacy
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Privacy-enhancing technologies
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Many deep learning approaches have been developed for EEG-based seizure detection; however, most rely on access to large centralized annotated datasets. In clinical practice, EEG data are often scarce, patient-specific distributed across institutions, and governed by strict privacy regulations that prohibit data pooling. As a result, creating usable AI-based seizure detection models remains challenging in real-world medical settings. To address these constraints, we propose a two-stage federated few-shot learning (FFSL) framework for personalized EEG-based seizure detection. The method is trained and evaluated on the TUH Event Corpus, which includes six EEG event classes. In Stage 1, a pretrained biosignal transformer (BIOT) is fine-tuned across non-IID simulated hospital sites using federated learning, enabling shared representation learning without centralizing EEG recordings. In Stage 2, federated few-shot personalization adapts the classifier to each patient using only five labeled EEG segments, retaining seizure-specific information while still benefiting from cross-site knowledge. Federated fine-tuning achieved a balanced accuracy of 0.43 (centralized: 0.52), Cohen's kappa of 0.42 (0.49), and weighted F1 of 0.69 (0.74). In the FFSL stage, client-specific models reached an average balanced accuracy of 0.77, Cohen's kappa of 0.62, and weighted F1 of 0.73 across four sites with heterogeneous event distributions. These results suggest that FFSL can support effective patient-adaptive seizure detection under realistic data-availability and privacy constraints.

Problem

• Detect epileptic seizures from EEG data with limited labeled examples
• Enable personalized models without centralizing sensitive patient data
• Overcome data scarcity and privacy constraints in clinical settings

Innovation

• Federated learning fine-tunes biosignal transformer across distributed sites
• Few-shot personalization adapts classifier using minimal patient-specific data
• Two-stage framework enables privacy-preserving seizure detection without data pooling