Pretraining on Sleep Data Improves non-Sleep Biosignal Tasks

2026-05-04

AI Summary

This study investigates whether multimodal biosignals from sleep can serve as effective pretraining data to enhance performance on non-sleep tasks, such as standard EEG and ECG applications. To this end, the authors propose a contrastive learning–based pretraining framework incorporating a leave-one-out target strategy, which learns generalizable representations from multiple heterogeneous sleep datasets and transfers them to eight downstream non-sleep EEG and ECG tasks. Experimental results demonstrate that the proposed approach consistently and significantly outperforms models trained from scratch across all tasks, with some achieving or even surpassing the performance of existing task-specific or foundation models. This work provides the first systematic evidence that pretraining on sleep data yields robust generalization capabilities for diverse non-sleep biosignal analysis tasks.

Abstract

Sleep foundation models have recently demonstrated strong performance on in-domain polysomnography tasks, including sleep staging, apnea detection, and disease risk prediction. In this work, we investigate whether sleep biosignals can serve as an effective pretraining distribution for learning representations that transfer beyond sleep to adjacent domains. Following sleep foundation models, we perform sleep-only multimodal contrastive pretraining (with a leave-one-out objective) and evaluate transfer to non-sleep EEG and ECG, two well-benchmarked biosignal modalities with heterogeneous datasets and clinically meaningful downstream tasks. Across eight downstream tasks spanning multiple EEG and ECG datasets, sleep pretraining consistently improves performance relative to training from scratch. Moreover, on several tasks, we achieve performance competitive with or surpassing prior specialized state-of-the-art and foundation models.

Problem

• sleep pretraining
• biosignal transfer
• non-sleep EEG
• non-sleep ECG
• foundation models

Innovation

• sleep pretraining
• biosignal transfer learning
• multimodal contrastive learning
• foundation models
• EEG/ECG representation