MIT-BIH Arrhythmia Database, version 1.0.0
George Moody and Roger Mark (2005). PhysioNet.
https://doi.org/10.13026/C2F305
https://physionet.org/content/mitdb/1.0.0/
Original publication: Moody GB, Mark RG (2001), The impact of the MIT-BIH Arrhythmia Database, IEEE Engineering in Medicine and Biology 20(3):45–50, https://doi.org/10.1109/51.932724
PhysioNet platform: Pollard et al. (2026). PhysioNet as a global platform for biomedical research. Nature Health. https://doi.org/10.1038/s44360-026-00096-z

Contains information from MIT-BIH Arrhythmia Database made available under the Open Data Commons Attribution License v1.0.
https://physionet.org/content/mitdb/view-license/1.0.0/
The extracted database retains this license. Provided as is, without warranties, under the license's sections 7–8.

Changes by aileelab: record 100, channel MLII, 0–300 seconds; conversion to mV with WFDB 4.3.1 and decimal JSON; native 360 Hz retained; no signal filtering/resampling. Original beat sample indices and annotation symbols retained. Demographics, medication and other header comments not republished. No synthetic biosignal data.
Reproduction: python/wearable/prepare_ecg.py. Source checksums: tests/wearable/reference.json.

PPG-DaLiA (2019), official UCI distribution
Attila Reiss, Ina Indlekofer, Philip Schmidt, Kristof Van Laerhoven.
https://archive.ics.uci.edu/dataset/495/ppg+dalia
https://doi.org/10.24432/C53890
Source archive: https://archive.ics.uci.edu/ml/machine-learning-databases/00495/data.zip
Documentation: https://archive.ics.uci.edu/ml/machine-learning-databases/00495/readme.pdf
Required publication attribution: Reiss A, Indlekofer I, Schmidt P, Van Laerhoven K (2019). Deep PPG: Large-Scale Heart Rate Estimation with Convolutional Neural Networks. Sensors 19(14):3079. https://doi.org/10.3390/s19143079

UCI distributes these data under Creative Commons Attribution 4.0 International (CC BY 4.0).
https://creativecommons.org/licenses/by/4.0/
https://creativecommons.org/licenses/by/4.0/legalcode
Retain attribution, license, source links and this modification notice on reuse. Offered as is, without warranties, under CC BY 4.0 section 5. No endorsement by the original authors is implied.
The authors' separate distribution page specifies scientific noncommercial use; this extract uses the UCI download with its explicit CC BY 4.0 grant. See evidence/wearable-data-terms.md for the recorded distinction. WESAD is not included.

Changes by aileelab: synchronized S1,S2,S3,S4,S5,S7 pickle records; activities Sitting (1), Stairs (2), Walking (7). For each person/activity, first occurrence +30 s rounded up to an even second, followed by a 120 s excerpt wholly within that activity. 18 excerpts, 2160 s total. Original wrist BVP 64 Hz, three-axis ACC 32 Hz in g, and wrist EDA 4 Hz in microsiemens. Chest EDA/EMG/Temp dummy channels excluded. BVP rounded to 2 decimal places; ACC/EDA to 6; no resampling or filtering in the export. ECG-derived HR labels rounded to 6 decimals and aligned to the same 8 s windows/2 s shift as the supplied documentation. No questionnaires, demographics or synthetic biosignals republished.
Reproduction: python/wearable/prepare_ppg.py. Exact boundaries and SHA256 checksums: evidence/wearable-ppg-manifest.json.

Mathematical typesetting: KaTeX, MIT License. See ../vendor/katex/LICENSE.