High precision ECG Database with annotated R peaks, recorded and filmed under realistic conditions
==================================================================================================

This database contains ECGs from 25 subjects. Each subject was recorded performing 5 different tasks for two minutes:
   * sitting
   * a maths test on a tablet
   * walking on a treadmill
   * running on a treadmill
   * using a hand bike

The following channels were recorded with two [Attys](https://www.attys.tech/) running synchronously:
   * Einthoven II and III with standard cables and the amplifier worn around the waist
   * Exercise cheststrap ECG which resembles approximtely V2-V1 with the ECG amplifier directly mounted on the strap
   * Acceleration in X/Y/Z whith the sensor mounted directly on the chest strap
   
The cheststrap ECG allowed R peak detection even while jogging at a very high precision (+/- one sample). The sampling rate was 250Hz at a resolution of 24 bits. The database contains the unfiltered, DC-coupled signals as originally recorded. In order to be able to link the ECG artefacts to the behaviour of the subject all but one subject gave permission to be filmed and the videos are also part of the database.


Data Format
-----------
 
The ECG recordings are stored in .tsv files (tab separated values) and have the following data format:

chest_strap_V2_V1  cables_Einth_II    cables_Einth_III	   acc_x   acc_y    acc_z 


Annotations
-----------

The sample locations of heartbeats have been annotated for both the chest_strap_V2_V1 and cables_Einth_II recordings, they are stored in annotation_cs.tsv and annotation_cables.tsv respectively. Where no annotation files exists, there was too much noise in the ECG recording to reliably annotate it. The annotations were generated first by an automatic detection algorithm and then visually checked and corrected so that every R peak is correct by the sample.


Videos
------

Where the participant has consented, there is a video for each of the tasks. The video and ECG data have been synchronised so they start and end at the same time.


Example Code
------------

There is a Python API which allows the database to be easily accessed. An example of how to use the API can be found in usage_example.py.