ECG-SDI: Single-lead Electrocardiogram Enables Continuous Sleep Depth Index Profiling and Scalable Digital Biomarkers for Sleep Health

Q: What is the clinical evidence from this study?

Study design: Cohort. Population: Sleep disturbance and obstructive sleep apnea (n=12392). Intervention: ECG-SDI-derived disturbed sleep subtype vs. Normal sleep subtype. Primary outcome: All-cause mortality in the SHHS cohort (fully adjusted Model 3) (HR 1.20, 95% CI 1.05-1.38, p=0.007).

Q: What does this research mean for the field?

Continuous sleep depth profiling via single-lead ECG identifies a disturbed sleep phenotype associated with increased all-cause mortality and sensitively tracks CPAP-induced improvements in sleep architecture. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

Donglin Xie; Qingshuo Zhao; Songchi Zhou; Haoyu Wang; Jingyu Wang; Yuxi Zhou; Thomas Penzel; Shenda Hong

doi:10.48550/arxiv.2610.23727

October 16, 2026Open Access

ECG-SDI: Single-lead Electrocardiogram Enables Continuous Sleep Depth Index Profiling and Scalable Digital Biomarkers for Sleep Health

View Full Paper

Ask AI

Bookmark

Share

Key result

ECG-SDI-derived disturbed sleep phenotyping is linked to ~20% higher all-cause mortality.

HR 1.20
95% CI 1.05-1.38
P=0.007
n=12,392

Why the study?

Conventional sleep staging provides a coarse-grained representation that obscures within-stage sleep depth variation.

Does a single-lead ECG deep learning framework (ECG-SDI) accurately profile continuous sleep depth, stratify mortality and clinical sleep risk, and track treatment response to CPAP across multiple polysomnography cohorts?

Study Design

Type

Cohort (n=12,392)

Multicenter

Yes

Structured PICO

Does a single-lead ECG deep learning framework (ECG-SDI) accurately profile continuous sleep depth, stratify mortality and clinical sleep risk, and track treatment response to CPAP across multiple polysomnography cohorts?

P

Population

12,392 overnight polysomnography recordings across multiple public sleep cohorts (SHHS, MESA, MrOS, CFS) and 507 paired pre/post-CPAP clinical patients in the BZ-CPAP cohort.

E

Exposure

ECG-SDI framework: a two-stage deep learning model (Net1D-based convolutional encoder and Transformer) applied to single-lead ECG (resampled to 100 Hz, 30-s epochs) to generate a continuous sleep depth index (0 to 1 scale) and digital sleep biomarkers; also evaluated pre- vs. post-continuous positive airway pressure (CPAP) therapy in the BZ-CPAP subcohort (n=507).

C

Comparator

Conventional polysomnography (PSG) discrete sleep-stage labels (Wake, N1, N2, N3, REM); normal sleep subtype vs. disturbed sleep subtype; pre-CPAP baseline vs. post-CPAP recordings.

O

Outcome

Correlation between ECG-SDI and ordered sleep stages excluding REM, and area under the receiver operating characteristic curve (AUROC) for REM classification.surrogate

Main Result

Hazard Ratio: 1.2 (95% CI 1.05–1.38)

p-value: p=0.007

A two-stage deep learning model applied to single-lead ECG accurately profiles continuous overnight sleep depth, identifies a disturbed sleep phenotype independently associated with elevated all-cause mortality, and captures improvements in sleep continuity following CPAP therapy.

Limitations

Retrospective multi-cohort study design
Research normalization and REM-related biomarkers (MDR, PR) in the public-cohort implementation relied on reference PSG sleep-stage labels
Lower sleep-depth correlations in the external BZ-CPAP cohort (0.456 pre-CPAP and 0.546 post-CPAP) compared to other cohorts
Wearable home workflow was only an illustrative proof-of-concept without home-based estimation accuracy or individual treatment-benefit validation
Broad wake distributions in SHHS and MESA potentially reflecting underlying cardiovascular burden
Wake distributions in SHHS and MESA were broader than expected with an upper tail extending to relatively high depth values, potentially reflecting cardiovascular burden and alterations in nocturnal cardiac dynamics.
Research normalization of depth scores for trajectory visualization and biomarker extraction used reference Wake/NREM sleep-stage labels (5th and 95th percentiles) in addition to the ECG-derived model output.
In the wearable proof-of-concept workflow, displayed trajectories and biomarker values were illustrative, and home-based estimation accuracy and individual treatment-monitoring performance were not evaluated.

Ask about this paper

Answers are grounded in this paper and cite their sources.

Implication

Sleep biomarker phenotyping may flag higher mortality risk; hypothesis-generating and should not yet change practice.

Abstract

Conventional sleep staging provides a coarse-grained representation that obscures within-stage depth variation. We developed ECG-SDI, a single-lead electrocardiography (ECG) framework for continuous sleep-depth profiling and digital biomarker extraction. Using 12,392 overnight polysomnographic recordings, we trained and validated a two-stage deep learning model supervised by conventional sleep-stage labels. Correlations with ordered stages excluding rapid eye movement (REM) sleep were 0.828 in the pooled internal test set and 0.848 in the external Cleveland Family Study; corresponding REM classification AUROCs were 0.954 and 0.977. Overnight trajectories captured within-stage variation. Biomarker-based phenotyping identified a disturbed subtype associated with poorer subjective sleep quality and higher all-cause mortality in the Sleep Heart Health Study (adjusted hazard ratio, 1.20; 95% CI, 1.05-1.38). Among 507 participants with paired recordings before and after continuous positive airway pressure (CPAP), normalized sleep-depth area increased by 6.1% and shallow-sleep burden below 0.3 decreased by 16.0%. Exploratory analyses linked reductions in approximate entropy and depth variability to reduced oxygen desaturation and increased sleep efficiency, respectively (Spearman rho = 0.24 and 0.12); both associations remained positive after covariate adjustment. Mean ECG-SDI within N2 increased by 0.017 (95% CI, 0.004-0.030). A wearable proof-of-concept illustrated the acquisition-to-report workflow. These findings support ECG-based sleep phenotyping and further evaluation of CPAP response monitoring at home.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Xie et al. (2026) conducted a cohort in Sleep disturbance and obstructive sleep apnea (n=12,392). ECG-SDI-derived disturbed sleep subtype vs. Normal sleep subtype was evaluated on All-cause mortality in the SHHS cohort (fully adjusted Model 3) (HR 1.20, 95% CI 1.05-1.38, p=0.007). ECG-SDI-derived disturbed sleep phenotyping was associated with higher all-cause mortality (adjusted HR 1.20, 95% CI 1.05-1.38) and captured CPAP-induced sleep depth improvements.

synapsesocial.com/papers/6ae28e75cbadf70daf6c42cc https://doi.org/10.48550/arxiv.2610.23727

View Full Paper

Ask AI

Bookmark

Share

ECG-SDI: Single-lead Electrocardiogram Enables Continuous Sleep Depth Index Profiling and Scalable Digital Biomarkers for Sleep Health

Study Design

Structured PICO

Main Result

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

ECG-SDI: Single-lead Electrocardiogram Enables Continuous Sleep Depth Index Profiling and Scalable Digital Biomarkers for Sleep Health

Study Design

Structured PICO

Main Result

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading