MIMIC-IV-ECG is a PhysioNet matched subset of about 800,000 ten-second 12-lead diagnostic ECGs from nearly 160,000 patients who appear in the MIMIC-IV Clinical Database. Records are sampled at 500 Hz in WFDB format and can be linked to MIMIC-IV clinical modules and, when available, cardiologist report identifiers. Released under the Open Data Commons Open Database License v1.0 (open access).
- License
- ODbL 1.0
- Type
- Dataset
- Access
- Not reported
- Subjects
- 160,000
- Signals
- ECG
- PPG sites
- None
- Setting
- Not reported
- Wrist
- No
- Raw waveforms
- Yes
- Tasks
- Fm pretraining
- Papers
- Updated
- Oct 27, 2026
Original description
The MIMIC-IV-ECG module contains approximately 800,000 diagnostic electrocardiograms across nearly 160,000 unique patients. These diagnostic ECGs use 12 leads and are 10 seconds in length. They are sampled at 500 Hz. This subset contains all of the ECGs for patients who appear in the MIMIC-IV Clinical Database. When a cardiologist report is available for a given ECG, we provide the needed information to link the waveform to the report. The patients in MIMIC-IV-ECG have been matched against the MIMIC-IV Clinical Database, making it possible to link to information across the MIMIC-IV modules.
Verbatim fromPhysioNet· Licence: Open Data Commons Open Database License v1.0

Key numbers
How it works
- Sampling
- 500 Hz (12-lead, 10 s records)
- Device roles
- Clinical reference (ECG)
Models trained or evaluated on it· 2
Papers using it· 6
- IECG: Weakly-Supervised Interpretability-Enhancing Foundation Model for ElectrocardiogramWACV 2027 · 2026WACV 2027 · 2026
ECG-JEPA: Learning General Representation of 12-Lead ECG with a Joint-Embedding Predictive ArchitectureNeural Networks · 2027Neural Networks · 2027- Scaling ECG Foundation Models and Identifying a Threshold for Effective Representation LearningmedRxiv preprint v1, posted 2026-07-17 · 2026medRxiv preprint v1, posted 2026-07-17 · 2026
MS-ECG-FM: Towards a More Universal Electrocardiogram Foundation Model for Health Monitoring using Multi-source Contrastive LearningarXiv preprint · 2026arXiv preprint · 2026- NeuroRVQ: Multi-Scale Biosignal Tokenization for Generative Foundation ModelsarXiv preprint · 2025arXiv preprint · 2025
QECGBERT: a scalable ECG foundation model based on self-supervised language learning via VQ-VAE waveform tokenizationarXiv 2610.33964v1 · 2026arXiv 2610.33964v1 · 2026
FAQ
How many subjects are in MIMIC-IV-ECG?
MIMIC-IV-ECG has data from 160,000 subjects, as listed in its research record.
What signals does MIMIC-IV-ECG include?
MIMIC-IV-ECG includes ECG data. Sampling rate as recorded: 500 Hz (12-lead, 10 s records). Raw waveforms are included.
What can MIMIC-IV-ECG be used for?
Research tasks it supports: foundation model pretraining.
How do I get MIMIC-IV-ECG, and what is its licence?
It is hosted on PhysioNet at https://physionet.org/content/mimic-iv-ecg/1.0/. Licence: ODbL 1.0.
Which models and papers in the ChatPPG research hub use MIMIC-IV-ECG?
The hub links it to models: IECG and QECGBERT (VQ-VAE beat-token BERT ECG FM; Afghah lab); 4 papers, including "ECG-JEPA: Learning General Representation of 12-Lead ECG with a Joint-Embedding Predictive Architecture". They are listed under Related on this page.
Cite
APA
MIMIC-IV-ECG [Data set]. (n.d.). physionet. https://doi.org/10.13026/4nqg-sb35
BibTeX
@misc{mimicivecgmimicivecg,
title = {{MIMIC-IV-ECG}},
howpublished = {physionet},
doi = {10.13026/4nqg-sb35}
}Built from the hub record. Check it against the publisher before submitting.
Cite the ChatPPG record
Cite the dated snapshot 2026-10 for a fixed, reproducible version. Please also cite the original publications listed above.
APA
ChatPPG. (2026). MIMIC-IV-ECG: dataset record (Version 2026-10) [Data set]. ChatPPG Research Hub. https://chatppg.com/research/datasets/mimic-iv-ecg
BibTeX
@misc{chatppg_dataset_mimic_iv_ecg,
author = {{ChatPPG}},
title = {{MIMIC-IV-ECG: dataset record}},
year = {2026},
month = oct,
version = {2026-10},
howpublished = {\url{https://chatppg.com/research/datasets/mimic-iv-ecg}},
note = {ChatPPG Research Hub, Snapshot 2026-10, https://chatppg.com/research/snapshots/2026-10}
}CITATION.cff
cff-version: 1.2.0 message: "If you use this resource, please cite it as below." type: dataset title: "MIMIC-IV-ECG: dataset record" authors: - name: "ChatPPG" url: "https://chatppg.com/research/datasets/mimic-iv-ecg" repository-artifact: "https://chatppg.com/research/snapshots/2026-10" version: "2026-10" date-released: "2026-10-27"
History
Record updated
Papers
ECG-JEPA: Learning General Representation of 12-Lead ECG with a Joint-Embedding Predictive ArchitectureUses MIMIC-IV-ECGNeural Networks · 2027
MS-ECG-FM: Towards a More Universal Electrocardiogram Foundation Model for Health Monitoring using Multi-source Contrastive LearningUses MIMIC-IV-ECGarXiv preprint · 2026
QECGBERT: a scalable ECG foundation model based on self-supervised language learning via VQ-VAE waveform tokenizationUses MIMIC-IV-ECGarXiv 2610.33964v1 · 2026
