ECG vs MS Band 2 dataset
Introduction
This dataset contains a small experiment designed to analyse the error present in Inter Beat Intervals (IBI) reported by the MS Band 2 wearable devices, using an ECG chest strap as ground truth.
This dataset contains PPG and ECG data of 4 subjects (1 additional subject is kept away from the dataset, to be used as test set), every subject was asked to:
- be as still as possible for 10 minute;
- climb up and down a ramp of stairs for 10 minutes;
- be as still as possible for 10 minutes;
- move freely for 10 minutes.
The sensors used are:
- Microsoft Band v2 (PPG and Accelerometer)
- RR data from a Polar H10
Data format
Data is organised in folders, one folder for each subject. Each folder contains 5 text files:
- ECG.csv: this file contains the RR intervals from the ECG chest strap, it contains 2 columns: timestamp and value. timestamp is the moment when the RR was received by the app from the band; value is the RR.
- ACC.csv: this file contains the accelerometer data from the Microsoft Band 2, it contains 4 columns: timestamp, x, y, z. timestamp is the moment when the data was received by the app; x, y, and z are the accelerometer data.
- PPG.csv: this file contains the PPG data from the MS Band 2, it contains 2 columns: timestamp and value. timestamp contains the moment in time at which the IBI was received by the app from the band; value contains the IBI duration. Important: the MS Band 2 sends data over bluetooth, and doesn't send the IBI immediately. As a consequence the timestamp should not be used as a strict indication of when the heartbeat happened, only approximately.
- sections.csv: this file contains 2 columns: offset and section. Offset contains the starting time of the section, in seconds, since the beginning of the experiment; section contains the name of the section.
- offset.csv: this file contains the estimated difference in time between timestamps from PPG and from ECG, useful to align the two time series.
References
For an analysis of the propagation of noise in PPG data from IBI to HRV features see http://digital-library.theiet.org/content/journals/10.1049/htl.2017.0039