ECG data was collected using a Actiwave Cardio sensor. This attaches to the chest using standard surgical ECG stick on connectors, as you would have in a hospital ECG device. Recordings are typically about 40 hours long going from morning one day to the evening the next including one night’s readings in between. As well as the ECG readings the device also contains a three axis accelerometer.The raw format for these files is EDF+ an industry standard binary format. However to make the data easier to analyse for those without specialised software, it has also been converted to CSV text files using the edf2ascii utility. Typically EDF files and zipped text files are 40-70 Mbytes each.
meta information, one line per file, as noted most files are over two days, but there may be short periods at the beginning and end where the sensor is not attached. Occasionally also the sensor was dislodged and had to be replaced leading to an anomalous inactive period.
fields:
date – a date stamp in format YYYY_MM_DD
start_date – \ ___ date and time when recording starts
start_time – /
number_of_records – always 8947776 = approx 40 hours x 64Hz samples
url – dropbox url of zipped CSV text file
edf_url – dropbox url of raw EDF+ file
identical data in JSON format
Each zipped CSV text file has the following format:
Time – timestamp in seconds
electrode — ECG in microVolts range +/- 9860uV
X — accelerometer in m/2^2
Y — “” “” “” “”
Z — “” “” “” “”
Note that the accelerometer readings are blank every second line as they are only sampled at 32Hz.
For the full EDF+ files see the EDF+ Specification.
extract:
Time,1,2,3,4 0.000000,-2158.635648,0.101894,-0.101894,10.147408 0.015625,-4306.157876,,, 0.031250,-2203.674246,0.101894,-0.101894,9.961602 0.046875,4488.359476,,, 0.062500,4483.680141,0.101894,-0.101894,9.961602 0.078125,-4308.497544,,, 0.093750,-4304.110667,0.101894,-0.101894,10.147408 0.109375,4487.774559,,, 0.125000,4483.387683,0.101894,-0.101894,10.147408
Hi, Alan – would be interested to look at your ECG data as a product I developed analyses holter ECG files.
However, none of the links on this page seem to work. Has the data been uploaded yet?
Sorry Iain, very belatedly catching up with bog comments 🙁
I have fixed the links, one was a file no uploaded, the over a mistyped URL :-/
see … very belated … reply to Quentin
Hi Alan,
Is there a way to get the Patient Id and the actual start time (full time stamp) into the ecg extract file (the one that has the time and voltage)?
Thanks,
Balaji
Dear Alan,
Basically I am a ECG interpretation technician, now I am facing a problem that one my client sent me some ECG file that are in excel format and he said it is .csv format. It is very new format to me. Can you guide me to open that file to view the ECG waveform and interpret it. Actually I don’t have better knowledge of MATLAB, or other programs but I have EDF browser in my computer. If you help me it will be more useful to me. Please help me to sort out this problem.
Hi Naresh
Although I collected the EDF data, I’m not an expert with EDF and biosensing formats. I did the EDF->CSV conversion using the tool that read the data from the ECG device.
However, did a quick peek and EDFBrowser an open source tool has an ASCII/CSV to EDF conversion mode.
There is also a standalone tool Ascii2Edf … with a very techie-weird website, but maybe useful if you have a lot of files to convert.
Alan
Dear Alan, I am very interested in your data for a research project, however, it seems that all dropbox links are now dead. Is there another place where these files are maintained ?
Thank you so much in advance for your reply !
sorry for taking so ling to get to this,. Dropbox changed their linking mechanisms which broke old public links 🙁
The EDF files are at: https://www.dropbox.com/sh/45zlfokaynpx3lr/AAD8yZvw3jbjHunCJX6194oSa?dl=0
and CSV versions at: https://www.dropbox.com/sh/idyvgodzdsgjspo/AAA3ujbX8MzvmzVWlIc6AoVXa?dl=0
I need to re-generate the meta files too … but will need to find the code I used to do that originally!