LGAINENCSep 3, 2025

StableSleep: Source-Free Test-Time Adaptation for Sleep Staging with Lightweight Safety Rails

arXiv:2509.02982v1h-index: 10
Originality Incremental advance
AI Analysis

This work addresses sleep staging accuracy for patients with unseen physiology or recording conditions, offering a practical solution for on-device or bedside use, though it is incremental in its adaptation approach.

The paper tackles the problem of sleep staging model degradation on unseen patient data by proposing a streaming, source-free test-time adaptation method with safety mechanisms, achieving consistent gains over a frozen baseline with per-stage metrics and Cohen's kappa reported.

Sleep staging models often degrade when deployed on patients with unseen physiology or recording conditions. We propose a streaming, source-free test-time adaptation (TTA) recipe that combines entropy minimization (Tent) with Batch-Norm statistic refresh and two safety rails: an entropy gate to pause adaptation on uncertain windows and an EMA-based reset to reel back drift. On Sleep-EDF Expanded, using single-lead EEG (Fpz-Cz, 100 Hz, 30s epochs; R&K to AASM mapping), we show consistent gains over a frozen baseline at seconds-level latency and minimal memory, reporting per-stage metrics and Cohen's k. The method is model-agnostic, requires no source data or patient calibration, and is practical for on-device or bedside use.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes