Evaluation of the Effectiveness of Post-filtration Smoothing using Lossless Compression for Heart Rate Variability Obtained from a Very Noisy ECG

Wearable devices have great potential in a personalized assessment of human health. In this regard, it seems, the most promising are devices that receive an ECG signal for further calculation of heart rate variability based on a sequence of RR intervals. However, the ECG signal is often exposed to high-intensity noise caused, in particular, by electrodes motion and muscle artefacts. The study shows the need to use the post-filtration smoothing, depending on the values of the ECG lossless compression ratio before and after filtration. Thresholds of ECG compression were found, below which it is recommended to use such smoothing. Also, noise types and heart rate values (HR) were determined at which smoothing improves the sensitivity and precision of determining R-peaks. Besides, a criterion was found, the use of which allows assessing the need for ECG smoothing to improve its quality, in terms of the mean square error (MSE), for all studied types of noise and HR values. This criterion is the difference between the lossless compression ratios of a smoothed ECG after filtering and the filtered ECG. If this difference is greater than zero, then the ECG quality will be improved through the use of smoothing, in terms of MSE. The results of the study are confirmed using the base of real ECGs - MIT-BIH Arrhythmia Database.