aiqbench analysis
This paper benchmarks frozen representations from self-supervised polysomnography foundation models against sleep-staging encoders pretrained at similar scale and modality coverage. Its main finding is that the apparent advantage of foundation models over a narrow EEG-only baseline disappears once supervised baselines are matched for scale and data, and that transfer from sleep scorers is more data-efficient. For ML readers, it is a useful reminder that fair pretraining comparisons in clinical time-series work require matched baselines.
On Sleep-Wake Disorder Identification from Polysomnographic Representations
arXiv:2610.17451v1 Announce Type: new Abstract: Self-supervised polysomnography foundation models (PSG FMs) aim to capture clinically useful information beyond sleep staging, yet are rarely compared with sleep-staging encoders pretrained at comparable scale and modality coverage. Using a common whole-night, missingness-aware state-space downstream modeling framework, we compared frozen representations from prominent large-scale, multi-cohort sleep-stage-supervised encoders and self-supervised PSG FMs for multilabel identification of six sleep-wake and circadian disorder groups in 7,906 PSGs from an unseen registry. Although PSG FMs outperformed an EEG-only, single-cohort supervised baseline, this ordering reversed when supervised pretraining approached comparable scale and modality coverage. Transfer learning from neural sleep scorers achieved higher overall performance while exhibiting greater downstream data efficiency. These findings underscore the need for scale- and modality-matched pretrained supervised baselines when assessing PSG FMs' downstream potential.
Abstract:Self-supervised polysomnography foundation models (PSG FMs) aim to capture clinically useful information beyond sleep staging, yet are rarely compared with sleep-staging encoders pretrained at comparable scale and modality coverage. Using a common whole-night, missingness-aware state-space downstream modeling framework, we compared frozen representations from prominent large-scale, multi-cohort sleep-stage-supervised encoders and self-supervised PSG FMs for multilabel identification of six sleep-wake and circadian disorder groups in 7,906 PSGs from an unseen registry. Although PSG FMs outperformed an EEG-only, single-cohort supervised baseline, this ordering reversed when supervised pretraining approached comparable scale and modality coverage. Transfer learning from neural sleep scorers achieved higher overall performance while exhibiting greater downstream data efficiency. These findings underscore the need for scale- and modality-matched pretrained supervised baselines when assessing PSG FMs' downstream potential.
Machine Learning (cs.LG)
arXiv:2610.17451 [cs.LG]
(or arXiv:2610.17451v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.17451 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 218 words
Related coverage
BeatGraph: Self-Supervised Heartbeat Graphs for Infant ECG Representations from the Home Environment
arXiv · Sep 28, 2026
DyssomniaNN: Multi-Scale Spiking Neural Modeling of Human Sleep Disorders and Perceived Restedness
arXiv · Oct 12, 2026
Learning transferable human physiology from two million hours of sleep with SleepFM-2
arXiv · Sep 9, 2026
Deep Learning Approaches for Sleep Apnea Classification from Polysomnographic EEG Signals
arXiv · Jul 20, 2026