--- pretty_name: Sleep Research Data tags: - sleep - actigraphy - polysomnography - physiological-signals --- # Sleep Research Data Public research mirrors of the original releases listed below. Each dataset retains its own license, attribution and source metadata; there is no blanket license covering all data. The owner explicitly authorized public hosting of license-cleared data on 19 November 2026 (Pacific/Auckland). ## Transfer status Only rows marked **Complete** represent a fully verified copy. A dataset directory or license file alone does not establish that all signals have been uploaded. See `inventory.json` for the verification timestamp and immutable data revision. | Dataset | Version | License | Status | | --- | --- | --- | --- | | [Sleep-EDF Expanded](https://physionet.org/content/sleep-edfx/1.0.0/) | 1.0.0 | ODC-By-1.0 | Complete; remote hashes verified | | [HMC Sleep Staging](https://physionet.org/content/hmc-sleep-staging/1.1/) | 1.1 | CC-BY-4.0 | Complete; remote hashes verified | | [CAP Sleep Database](https://physionet.org/content/capslpdb/1.0.0/) | 1.0.0 | ODC-By-1.0 | Complete; remote hashes verified | | [PhysioNet Challenge 2018](https://physionet.org/content/challenge-2018/1.0.0/) | 1.0.0 | ODC-By-1.0 | Complete; remote hashes verified | | [MIT-BIH Polysomnographic](https://physionet.org/content/slpdb/1.0.0/) | 1.0.0 | ODC-By-1.0 | Complete data; 113/114 source files (legacy webpage unavailable) | | [SleepAccel (PhysioNet release)](https://physionet.org/content/sleep-accel/1.0.0/) | 1.0.0 | ODC-By-1.0 | Complete; remote hashes verified | | [BIDSleep](https://physionet.org/content/bidsleep-dataset/1.0.1/) | 1.0.1 | ODC-By-1.0 | Download staged or partial; not uploaded | | [Aalborg University Wearable Sleep Study (AAUWSS)](https://zenodo.org/records/16919071) | 1.1 | cc-by-4.0 | Complete; remote hashes verified | | [Newcastle polysomnography and accelerometer data](https://zenodo.org/records/1160410) | 1.0 | cc-by-4.0 | Download staged or partial; not uploaded | | [PSGWatch](https://doi.org/10.58132/DCQYO1) | 1.0 | CC-BY-4.0 | Complete; remote hashes verified | | [ISRUC-Sleep (NeMAR BIDS conversion)](https://nemar.org/dataset/nm000111) | 1.0.1 | n/a | Held: redistribution terms need clarification | | [Wearanize+ OA](https://data.ru.nl/collections/di/dcmn/DSC_wrnzpoa_t0000925a_195) | 1 | https://creativecommons.org/licenses/by/4.0/legalcode.en | Held: redistribution terms need clarification | ## Layout and integrity Each dataset folder contains its original license, citation/source information and provenance. Original archives remain unchanged; per-file mirrors preserve source paths under `files/`. PhysioNet files are checked against the official SHA256 manifests; Zenodo and Dataverse releases are checked against their published archive/file checksums. Remote content hashes are verified after each commit. Local staging copies may then be removed without deleting the HF copy or verification records. Sleep-EDF's 398 original files were transferred from a pinned HF mirror only after every SHA256 matched PhysioNet. No processed arrays from that mirror were copied. Challenge 2018 uses its full raw training/test files, not the small website ZIP; accompanying research papers and example-analysis code are outside this data mirror's scope. ## Data distinctions - SleepAccel and BIDSleep provide heart-rate and motion measurements with sleep labels; heart rate is not raw PPG and labels are not raw PSG. - AAUWSS includes paired raw PSG and wearable BVP/PPG/ACC. Its aligned pickle files were not deserialized during verification. Do not load untrusted pickles solely to inspect data. - Newcastle contains wrist acceleration and PSG-derived sleep scores, not raw PSG. Use filename participant IDs, not incorrect IDs inside accelerometer headers. - PSGWatch includes 43 raw PSG recordings and 37 scored recordings. Its actigraphy ZIP is only one component. - ISRUC and Wearanize+ are not uploaded while redistribution terms remain unresolved. Preserve the original source notices and cite the contributing researchers. Do not attempt to identify participants. These are research data, not a clinical diagnostic tool. Public hosting does not remove any source-license conditions.