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

TL;DR

This paper introduces a federated few-shot learning framework for EEG seizure detection that respects privacy constraints and adapts to individual patients with minimal labeled data, achieving promising accuracy improvements.

Contribution

The paper presents a novel two-stage federated few-shot learning approach for personalized EEG seizure detection, combining federated training of a transformer model with patient-specific fine-tuning.

Findings

Federated fine-tuning achieved balanced accuracy of 0.43, Cohen's kappa of 0.42, and F1 of 0.69.

Client-specific models in FFSL reached an average balanced accuracy of 0.77.

The approach supports effective patient-adaptive seizure detection under privacy constraints.

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.