BeatGraph: Self-Supervised Heartbeat Graphs for Infant ECG Representations from the Home Environment

2026-09-25

AI Summary

This study addresses the limitations of existing ECG foundation models, which often overlook cardiac structural information and struggle to accommodate infant tachycardia and home-based monitoring scenarios. We propose a novel graph-based modeling approach that treats individual heartbeats as nodes, constructing heartbeat graphs from 30-second windows. This framework integrates a residual graph attention network, shared encoders, and Transformer-based temporal modeling, trained via masked prediction self-supervised pretraining for multi-task learning. Additionally, we release the first public in-home infant ECG dataset annotated with physiological states and emotional labels. Experimental results demonstrate that our method significantly improves performance across multiple tasks, exhibits strong cross-age transferability, and achieves state-of-the-art results on adult benchmarks.

Abstract

Electrocardiogram (ECG) foundation models typically tokenize the signal into fixed-length patches that ignore cardiac structure, so a patch may split a heartbeat and the number of beats in each patch shifts with heart rate. This matters most for infants, whose heart rates are higher and whose ECG differs from the adult, clinic-recorded 12-lead data these models are built on. A model for infant ECG should therefore reason about heartbeats directly rather than recover them from arbitrary patches. We propose BeatGraph, which makes the heartbeat its unit of representation, modeling each 30-second window as a graph of beats. A shared beat encoder embeds each heartbeat from its waveform and inter-beat intervals, a Transformer with positional encoding orders the beats in time, and residual graph attention layers relate every beat to every other before attention pooling yields a window embedding. We pretrain BeatGraph on our new corpus of unlabeled infant recordings by predicting masked-beat embeddings, then fine-tune it for each task. One backbone supports sleep-wake detection, infant-state classification, activity-source identification (infant- or caregiver-initiated movement), and affect recognition, improving macro-F1 over the strongest baseline on each task by 0.076 to 0.158. It also transfers across age groups, reaching 0.892 AUROC on the ZZU-pECG pediatric benchmark (ages 0 to 14), within 0.001 of the best published self-supervised ECG model, and matching that model under linear evaluation on the adult PTB-XL benchmark despite infant-only pretraining. Finally, to our knowledge, we release the first public infant ECG corpus collected in homes, classrooms, and laboratory settings with state and affect labels. It contains 3,408 hours of single-channel ECG from 143 infants aged 3 to 11 months, with unlabeled pretraining data, benchmark tasks, and subject-level splits.

Problem

• Infant ECG
• Self-supervised learning
• Heartbeat representation
• Foundation models
• Home environment

Innovation

• Self-supervised learning
• Heartbeat graph representation
• Infant ECG
• Graph attention network
• Masked beat prediction