Characterization and reduction of exercise-based motion influence on heart rate variability using accelerator signals and channel decoding in the time-frequency domain

• Published in: Physiological measurement Vol. 39; no. 11; pp. 115002 - 115012
• Main Authors: Alikhani, Iman, Noponen, Kai, Hautala, Arto, Seppänen, Tapio
• Format: Journal Article
• Language: English
• Published: England IOP Publishing 30.10.2018
• Subjects: electrocardiography; exercise; heart rate variability; motion attenuation; motion influence; time-frequency representation
• ISSN: 0967-3334; 1361-6579; 1361-6579
• DOI: 10.1088/1361-6579/aadeff

Objective: Heart rate variability (HRV) is defined as the variation of the heart's beat to beat time intervals. Although HRV has been studied for decades, its response to stress tests and off-rest measurements is still under investigation. In this paper, we studied the influence of motion on HRV throughout different exercise tests, including a maximal running of healthy recreational runners, cycling, and walking tests of healthy subjects. Approach: In our proposed method, we utilized the motion trajectory (which is known to exist partially in HRV) measured by a three-channel accelerator (ACC). We then estimated their shares in HRV using a wearable electrocardiogram (ECG) and an error-correcting problem formulation. In this method, we characterized the motion components of three orthogonal directions induced into the HRV signal, and then we suppressed the estimated motion artefact to construct a motion-attenuated spectrogram. Main results and Significance: Our analysis showed that HRV in the exercise context is susceptible to motion artefacts. Furthermore, the interpretation of autonomic nervous system (ANS) activity and HRV indices throughout exercise has a high margin of error depending on the intensity level, type of exercise, and motion trajectory. Our experiment on 84 healthy subjects throughout mid-intensity cycling and walking tests showed 39% and 32% influence on average, respectively. In addition, our proposed method revealed through a maximal running test with 11 runners that motion can describe on average 20%-40% of the HRV high-frequency (HF) energy at different workloads of running.