[2603.00190] OSF: On Pre-training and Scaling of Sleep Foundation Models

Computer Science > Machine Learning arXiv:2603.00190 (cs) [Submitted on 27 Feb 2026] Title:OSF: On Pre-training and Scaling of Sleep Foundation Models Authors:Zitao Shuai, Zongzhe Xu, David Yang, Wei Wang, Yuzhe Yang View a PDF of the paper titled OSF: On Pre-training and Scaling of Sleep Foundation Models, by Zitao Shuai and 4 other authors View PDF HTML (experimental) Abstract:Polysomnography (PSG) provides the gold standard for sleep assessment but suffers from substantial heterogeneity across recording devices and cohorts. There have been growing efforts to build general-purpose foundation models (FMs) for sleep physiology, but lack an in-depth understanding of the pre-training process and scaling patterns that lead to more generalizable sleep FMs. To fill this gap, we curate a massive corpus of 166,500 hours of sleep recordings from nine public sources and establish SleepBench, a comprehensive, fully open-source benchmark. Leveraging SleepBench, we systematically evaluate four families of self-supervised pre-training objectives and uncover three critical findings: (1) existing FMs fail to generalize to missing channels at inference; (2) channel-invariant feature learning is essential for pre-training; and (3) scaling sample size, model capacity, and multi-source data mixture consistently improves downstream this http URL an enhanced pre-training and scaling recipe, we introduce OSF, a family of sleep FMs that achieves state-of-the-art performance across nine datasets ...