README_Dataset

Title
12-Lead ECG Dataset for Cognitive Stress Detection Using HRV and ECG Morphology Features

Repository folder name
ECG_Stress_Dataset

DOI
10.5281/zenodo.19864426

Repository URL
https://doi.org/10.5281/zenodo.19864426

Authors / Creators
Salvador Ortiz-Santos
Georgina Mota-Valtierra
Jesús-Norberto Guerrero-Tavares
Xóchitl Siordia-Vásquez
Miguel Rojas-Hernández
Juvenal Rodríguez-Reséndiz

Recommended citation
Ortiz-Santos, S.; Mota-Valtierra, G.; Guerrero-Tavares, J.-N.; Siordia-Vásquez, X.; Rojas-Hernández, M.; Rodríguez-Reséndiz, J. 12-Lead ECG Dataset for Cognitive Stress Detection Using HRV and ECG Morphology Features. Zenodo, 2026. https://doi.org/10.5281/zenodo.19864426

Dataset description
This repository contains de-identified 12-lead electrocardiographic (ECG) recordings acquired from healthy young adult participants under baseline and cognitive stress conditions.

The dataset is shared as raw ECG recordings as exported from the TLC 6000 Holter system. The signals included in this repository have not been filtered, centered, segmented, cropped, or preprocessed. Therefore, the files represent the acquisition stage before the signal processing procedures reported in the associated manuscript.

The dataset was generated as part of a biomedical engineering study focused on ECG-based cognitive stress detection using heart rate variability (HRV), ECG morphology features, adaptive genetic feature selection, and multi-classifier evaluation.

Signal acquisition
ECG recordings were acquired using a TLC 6000 Holter system. The shared files preserve the original signal structure obtained from the device export. The recordings correspond to 12-lead ECG data, including the following leads:

I, II, III, aVR, aVL, aVF, V1, V2, V3, V4, V5, and V6.

The signals are provided before the preprocessing stage. Users who reuse this dataset should apply their own preprocessing pipeline according to the objective of their analysis. In the associated manuscript, preprocessing included robust centering, robust outlier clipping, high-pass filtering, low-pass filtering, and final centering before feature extraction.

Experimental protocol
The acquisition protocol included a baseline relaxation condition and a cognitive stress condition. Cognitive stress was induced using a structured task based on the Factor R subtest of the PMA-R battery.

The general experimental procedure included:
- A pre-rest adaptation phase before ECG recording.
- A baseline relaxation phase.
- A brief instruction phase before the cognitive stress task.
- A cognitive stress phase with ECG recording.
- A recovery period after task completion.

Although the associated manuscript analyzed selected ECG segments after preprocessing and segment selection, the files shared in this repository correspond to the raw Holter exports prior to filtering and cropping.

Participants
The associated study included healthy Biomedical Engineering students aged 18 to 25 years. Participants were instructed to sleep adequately, avoid physical exercise during the previous 12 hours, report no history of cardiac problems, refrain from coffee or energy drink consumption, and hydrate before acquisition. Participants with relevant physical discomfort or stress-related headache at the time of acquisition were excluded.

Ethics statement
The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the Faculty of Engineering of the Autonomous University of Querétaro under approval number CEAIFI-183-2021-PI. Written informed consent was obtained from all participants before data acquisition.

De-identification and privacy
The shared files were de-identified before publication. Participant names and direct personal identifiers were removed and replaced with coded identifiers such as ID-001, ID-002, ID-003, etc.

The private correspondence table linking coded IDs with original participant names is not included in this repository and must not be shared publicly.

Data organization
The dataset is organized in the folder:

ECG_Stress_Dataset/

The files are named using coded identifiers:

ID-001
ID-002
ID-003
...
ID-XXX

Each coded file corresponds to a de-identified ECG recording exported from the TLC 6000 Holter system. No Excel file or participant-identification table is included in this repository.

Suggested repository structure
ECG_Stress_Dataset/
|
|-- README_Dataset.txt
|
`-- ECG_Data/
    |-- ID-001
    |-- ID-002
    |-- ID-003
    `-- ...

Important notes for reuse
- The signals are raw and have not been filtered.
- The signals are not centered.
- The signals are not cropped or segmented.
- The files preserve the original acquisition/export condition from the TLC 6000 Holter system.
- Users should verify the signal format and apply an appropriate preprocessing workflow before analysis.
- The repository does not include the private ID-to-name correspondence table.
- The repository does not include informed consent forms or direct personal identifiers.

Suggested metadata to report when reusing the dataset
When using this dataset, researchers should report:
- The ECG leads analyzed.
- The preprocessing filters applied.
- The segment duration selected for analysis.
- The method used for R-peak detection or QRS detection.
- The HRV and/or morphology features extracted.
- The classifier or statistical model used.
- The validation strategy used for performance assessment.

Data use
This dataset is provided to support transparency, reproducibility, and further research in:
- ECG-based cognitive stress detection
- Heart rate variability analysis
- ECG morphology analysis
- Biomedical signal processing
- Feature selection
- Machine learning classification

Limitations
The dataset was acquired from healthy young adult participants under controlled experimental conditions. Therefore, results derived from these data should be interpreted with caution when generalizing to broader populations, clinical groups, older adults, or real-world stress scenarios.

Because the repository contains raw Holter exports, users are responsible for applying adequate preprocessing, quality control, artifact rejection, and segment selection before conducting HRV, ECG morphology, or machine learning analyses.

Data availability statement
The original raw ECG recordings presented in the associated study are openly available in Zenodo at https://doi.org/10.5281/zenodo.19864426. The dataset has been de-identified to protect participant privacy.

License
[Insert selected Zenodo license, e.g., Creative Commons Attribution 4.0 International (CC BY 4.0), if appropriate.]

Contact
For questions regarding the dataset, please contact the corresponding author(s):

Georgina Mota-Valtierra
Universidad Autónoma de Querétaro
Email: georgina.mota@uaq.mx

Juvenal Rodríguez-Reséndiz
Universidad Autónoma de Querétaro
Email: juvenal@uaq.edu.mx

Version
Version 1.0
Publication date: 2026-04-28
