Key result
LeadWeaver introduces a topology-aware framework to improve ECG classification under missing-lead conditions.
Why the study?
Most neural ECG classifiers assume all twelve leads are present, correctly ordered, and acquired under identical device conditions, rather than explicitly modeling inter-lead topology and uncertainty under missing, corrupted, or device-shifted leads.
Does LeadWeaver (topology-aware masked-lead modeling) improve multi-label ECG diagnostic accuracy and calibration under missing, corrupted, or device-shifted leads compared to standard neural ECG classifiers and missing-lead baselines in clinical ECG datasets?
Does LeadWeaver (topology-aware masked-lead modeling) improve multi-label ECG diagnostic accuracy and calibration under missing, corrupted, or device-shifted leads compared to standard neural ECG classifiers and missing-lead baselines in clinical ECG datasets?
Answers are grounded in this paper and cite their sources.
This preprint proposes LeadWeaver, a topology-aware machine learning framework designed to maintain diagnostic accuracy and calibrated uncertainty when interpreting 12-lead ECGs with missing or corrupted leads, though the reported quantitative results in the tables are explicitly noted as illustrative placeholders.
Clinical electrocardiography is organized as a structured set of correlated leads, yet most neural ECG classifiers assume that all twelve leads are present, correctly ordered, and acquired under the same device conditions at training and deployment. We propose LeadWeaver, a topology-aware ECG representation model designed for diagnosis under missing, corrupted, or device-shifted leads. LeadWeaver represents each lead as a temporal token stream, couples leads through a fixed-plus-learned physiological lead graph, and performs masked-lead latent completion rather than waveform synthesis. A reliability router estimates which observed leads should dominate each diagnostic decision, while a consistency objective aligns complete-lead and incomplete-lead predictions. We evaluate the model on PTB-XL and CPSC-style settings using complete-input diagnosis, arbitrary missing-lead subsets, lead corruption, cross-dataset transfer, calibration, and selective prediction. The central hypothesis is that robust ECG diagnosis requires explicitly modeling both inter-lead topology and uncertainty about unavailable evidence, rather than treating missing channels as zero-valued inputs.
No takes yet. Share an insight, caveat, or question.
Schneider et al. (2026) studied 12-lead ECG arrhythmia and morphology classification under missing or corrupted leads. LeadWeaver (topology-aware masked-lead ECG representation model) vs. 1D ResNet, CNN-BiLSTM, Transformer, State-space encoder, zero filling, mask tokens, channel dropout, deterministic completion, and waveform reconstruction was evaluated on Macro AUROC, macro AUPRC, macro F1, and calibration (ECE, Brier score) across complete and missing-lead ECG subsets. LeadWeaver proposes a topology-aware masked-lead ECG framework combining a physiological lead graph, probabilistic latent completion, and reliability routing (tables contain illustrative placeholders).
Synapse has enriched 10 related papers. Consider them for comparative context: