E4 data - BVP expected signal

Updated:

The BVP is the Blood Volume Pulse, and it's the primary output from the PPG sensor. The function it serves is explained here: Utilizing the PPG/BVP signal.

The BVP signal is provided as a .csv file that can be downloaded from our E4 Connect portal. We don't provide a unit of measurement for BVP because it's derived from the combination of two different measures of the amount of light that is reflected and goes back to the sensor.

The PPG/BVP is the input signal to the algorithms that compute inter-beat-interval (IBI) timings and heart rate (HR) as outputs:

  • The inter-beat-interval (IBI) is computed by detecting peaks (beats) of the BVP and computing the lengths of the intervals between adjacent beats. Further information can be found here: E4 data - IBI expected signal.

The IBI and HR outputs are provided as .csv files, that can be downloaded from our E4 Connect portal.

The BVP signal

The BVP signal is obtained from the PPG sensor by a proprietary algorithm which combines the light signals observed during both green and red exposure. It has a fixed sampling rate of 64 Hz (64 times per second).
In the figure below there is an example of a PPG signal obtained in a typical condition. The output may vary depending on the user and environmental settings.

blobid0.png

The significant points are presented below:

  • Diastolic points: local minima of the PPG used to compute the IBI.
  • Systolic points: local maxima of the PPG could be used in conjunction with the Diastolic point to estimate the vasoconstriction of the subject.
  • Dicrotic notch: under some conditions (i.e subject static) we observe the presence of this point which can be used to study different types of cardiac diseases.
  • Dicrotic wave: effect of Dicrotic notch


Expected output

The device must be worn snugly (but not uncomfortably tight) to receive good quality data.
Below you can find some examples of good and coherent BVP data output. These session outputs are from the E4 Connect portal.

Example of a session with good BVP output:

blobid1.png

 

Example of the signal at the beginning of the session (the first 30 seconds)

blobid2.png

 

Example of the signal when it has stabilized (some seconds after the beginning of the recording).

blobid3.png

 

Some notes on the signal:

  • Zero mean value.
  • The most important features of this signal are the frequency and the shape of the BVP signal, and not its intensity.
  • The signal represented on the E4 Connect portal is inverted in sign with respect to the classic BVP signal, therefore the light observed at a systolic point is less than the one observed at a diastolic point.
  • The difference BVP(Systolic point)-BVP(Diastolic point) is basically the difference in light between the most and least oxygenated blood condition. If the sensor is well positioned and there is no influence of motion on the data, this difference in light provides information on the vasoconstriction of the subject.
  • The BVP visualized on the E4 Connect portal is scaled, undersampled, and inverted with respect to the raw BVP signal that can be downloaded and that is contained in the BVP.csv file. The representation on the Connect is reliable anyway since the distance between peaks and waveform isn’t modified.
  • The downloaded BVP.csv file is the reference signal and can usually assume values in the range of [-500, 500]. The first row of the file contains the UNIX timestamp of the beginning of the session. In the 2nd row, you’ll find the sampling frequency that is fixed at 64 Hz. From the 3rd row onwards, there are the samples equally spaced in time (one every 1/64 sec).
  • An example of data from the session above imported and plotted in Matlab.blobid4.png
  • An example of the extract of the BVP.csv file.

Screenshot_2019-06-28_at_17.43.39.png

 

 

Have more questions?