ECG, Heart rate and heart rate variability
The following algorithms are based on single channel ECG signals that are acquried by the movisens sensors ekgMove, EcgMove 3, or EcgMove 4.
The following process is used to analyse ECG signals and to calculate Heart Rate Variability parameters.
- Artefact detection based on raw ECG signal: The ECG signal amplitude and the number of zero crossings per second have to be within a normal physiological range.
- R-Peak detection: R peaks are detected with an adapted algorithm from Hamilton, 2002.
- NN-List generation: The R-Peaks are filtered by an algorithm that was adapted from Clifford et al. 2002. The signal is checked for valid changes of consecutive RR-intervals, and also for valid changes of consecutive R-peak amplitudes. Most of the detected R-peaks and RR-intervals respectively that do not relate to nromal heart beats are filtered out. The final list of beats is the NN-List.
- Segmentation: The RR interval list is segmented into 2 minute segments with a shift of 30 seconds
- Segment validation: Each segment is checked if enough NN intervals are available for further analysis. Segments with insufficient NN intervals are discarded.
- Heart Rate calculation
- Detrending: In ambulatory settings stationarity of the heart rate is main concern for heart variability calculation. Therefore the NN-intervals are detrended to meet stationarity criteria Eleuteri et al. 2012
- Calculation of statistical HRV parameters: The parameters SDNN, RMSSD, SDSD, SD1, SD2, pNN50 and Baevsky SI are calculated.
- Interpolation: The NN-List is resampled with 4Hz
- Calculation of frequency domain HRV parameters: The power spectral density is estimatied using Welch’s method. The the parameters LF, HF, LF/HF parameters are calculated from the spectrum.
Notes for the calculation of HRV parameters (internal segmentation vs. output intervals):
- All HRV calculations occur internally every 30 seconds over segments of 2 minutes duration. For each output interval the mean values of the internally calculated values are output.
- If the output interval is smaller than 30 seconds, values are repeated.
- Hamilton, Patrick S. "Open source ECG analysis software documentation." Computers in cardiology 2002 (2002): 101-104
- Clifford, G. D., P. E. McSharry, and L. Tarassenko. "Characterizing artefact in the normal human 24-hour RR time series to aid identification and artificial replication of circadian variations in human beat to beat heart rate using a simple threshold." Computers in cardiology. IEEE, 2002
- Eleuteri, Antonio, et al. "An efficient time-varying filter for detrending and bandwidth limiting the heart rate variability tachogram without resampling: MATLAB open-source code and internet web-based implementation." Computational and mathematical methods in medicine 2012 (2012).
- Electrophysiology, Task Force of the European Society of Cardiology the North American Society of Pacing. "Heart rate variability: standards of measurement, physiological interpretation, and clinical use." Circulation 93.5 (1996): 1043-1065.
This parameter displays the exact positions of the R peaks of the ECG as events. The temporal resolution depends on the sample rate of the ECG. The default samplerate of movisens ECG sensors is 1024Hz.
The following figure shows a 10 second snapshot of the raw ECG signal and the R peaks underneath.
Nornal-to-normal intervals (Nn)
This parameter calculates the positions of the normal beats and the distance in [ms] to the last normal beat. It will show one value per normal heartbeat. A graph showing the RR or NN intervals as time series is also referred to as tachogramm.
movisensLIVE: StepCount can be calculated live on the sensor.
Beat by beat heart rate (HrBxB)
The heart rate values are output beat by beat, with one value per beat being displayed. The unit is [1/min].
Heart Rate (Hr)
This parameter caluclates the mean heart rate per output interval. The unit is [bpm].
movisensLIVE: Hr can be calculated live on the sensor.
High Frequncy HF (HrvHf)
This parameter calculates the mean value of the HRV parameter High Frequency (HF) per output interval. HF describes the spectral power in the frequency band between 0.15 and 0.4 Hz. The unit is [ms2]. The output interval is adjustable – see the top of this section for more details.
Low Frequncy lF (HrvLf)
This parameter calculates the mean value of the HRV parameter Low Frequency (LF) for each output interval. LF describes the spectral power in the frequency band between 0.04 and 0.15 Hz. The unit is [ms2]. The output interval is adjustable – see the top of this section for more details.
HRV parameter Low to High Frequency Ratio LF/HF (HrvLfHf)
HRV parameter Low to High Frequency Ratio (LF/HF) This parameter calculates the mean value of the HRV parameter Low Frequency to High Frequency Ratio (LF/HF) per output interval.
HRV parameter RMSSD (HrvRmssd)
This parameter calculates the mean value of the HRV parameter RMSSD per output interval. RMSSD is the root mean square of successive differences of beat intervals. The unit is [ms]. The output interval is adjustable – see the top of this section for more details.
movisensLIVE: HrvRmssd can be calculated live on the sensor.
HRV parameter pNN50 (HrvPnn50)
This parameter calculates the mean value of the HRV parameter pNN50 per output interval. pNN50 is the percentage of NN intervals greater than 50ms. The unit is [%]. The output interval is adjustable – see the top of this section for more details.
HRV parameter SD1, SD2, SD1/SD2 (HrvSd1, HrvSd2, HrvSd2Sd1)
SD1 and SD2 represent short and long term variability repectively. SD1 and SD2 can be derived from the Poincaré plot of the beat intervals. SD1 is the short half axis, SD2 is the long half axis of a fitted ellipse. SD1 and SD have the unit [ms]. Also the ratio SD2/SD1 can be calculated. The unit is . The output interval is adjustable – see the top of this section for more details.
This parameter calculates the mean value of the HRV parameter SDNN per output interval. SDNN describes the standard deviation of the beat intervals. The unit is [ms]. The output interval is adjustable – see the top of this section for more details.
Beavsky Stress Index (BaevskStressIndex)
This parameter was developed by R.M. Baevsky. It is based on frequency distribution of the heart beat intervals. It has no unit . The output interval is adjustable – see the top of this section for more details.
- Baevsky, R.M. Die Methodik der Analyse der Herzrhythmusvariabilität (übersetzt), 1999
- Baevsky, R M. Noninvasive methods in space cardiology. Journal of cardiovascular diagnosis and procedures. 1997, Bd. 14, 503-51
- Baevsky, R M und andere. HRV Analysis under the usage of different electrocardiography systems. s.l.: Committee of Clinic Diagnostic Apparatus and the Committee of New Medical Techniques of Ministry of Health of Russia. 2004.
ECG derived respiration (Edr)
Respiration can be derived from the ECG signal when the participant is at rest. A synthetic respiration signal is created from the amplitudes of the R peaks in the ECG, and from this signal the respiration rate can be calculated. The respiration rate might be distorted when the ECG signal contains artefacts e.g. due to movement. Beside the ECG signal, acceleration signal is required to calculate this parameter.
List of inter beat intervals as text file
This generates the text file ‘ReportIbi.txt’ that contains all RR intervals (time between the current and last R peak). Each line of the text file shows one value as output. The unit is [ms].
|Report||Type||Description||Aligned at full days|
|ReportHrvPdf||Detailed report with plots of activity class, MET, heart rate, HRV spectrogram, LF, HF, LF/HF, Baevskii Stress Index, Table of HRV parameters in different activity classes (one page per day). Preview.||no|