BiTimeCrossNet: Time-Aware Self-Supervised Learning for Pediatric Sleep

TL;DR

BiTimeCrossNet is a novel self-supervised learning framework that leverages temporal information and cross-modal interactions to improve pediatric sleep analysis across multiple tasks.

Contribution

Introduces BTCNet, a time-aware multimodal self-supervised model that captures segment timing and physiological signal interactions without task labels.

Findings

Outperforms non-time-aware variants in sleep-related tasks

Achieves strong results on respiration detection

Generalizes well to independent pediatric datasets

Abstract

We present BiTimeCrossNet (BTCNet), a multimodal self-supervised learning framework for long physiological recordings such as overnight sleep studies. While many existing approaches train on short segments treated as independent samples, BTCNet incorporates information about when each segment occurs within its parent recording, for example within a sleep session. BTCNet further learns pairwise interactions between physiological signals via cross-attention, without requiring task labels or sequence-level supervision. We evaluate BTCNet on pediatric sleep data across six downstream tasks, including sleep staging, arousal detection, and respiratory event detection. Under frozen-backbone linear probing, BTCNet consistently outperforms an otherwise identical non-time-aware variant, with gains that generalize to an independent pediatric dataset. Compared to existing multimodal self-supervised sleep models, BTCNet achieves strong performance, particularly on respiration-related tasks.