Abstract / Summary
Ambulatory Holter electrocardiogram (ECG) monitoring is increasingly used in ambulatory elderly populations; however, age-related physiological and mechanical factors substantially increase susceptibility to motion-induced signal artifacts, limiting reliable heart rate (HR) variability (HRV) analysis. This study aims to quantify age-group differences in motion artifact burden and to establish an age-stratified noise model using simultaneous ECG and three-axis accelerometry. Twenty participants—15 elderly patients (elderly male: OM, n = 10; elderly female: OF, n = 5) and five young males (young male: YM, n = 5)—underwent ambulatory ECG recording during daily life activities. The simultaneous acquisition of three-channel ECG and triaxial acceleration data enables quantitative characterization of motion-induced noise in the R–R interval (RRI) time series. We present a data processing pipeline that (1) extracts RRI sequences and computes a comprehensive set of heart rate variability (HRV) metrics (time-domain, frequency-domain, and nonlinear), (2) computes per-epoch accelerometer root mean square (RMS) as a motion intensity proxy, and (3) correlates motion intensity with RRI artifact rate across age groups. Key findings: Artifact rate was highest in OM (1.9 ± 4.7%) versus OF (0.9 ± 0.9%) and YM (0.3 ± 0.6%); standard deviation (SD) of all normal-to-normal intervals (SDNN) was 72.7 ± 26.5 ms (OM), 95.8 ± 71.8 ms (OF), and 79.6 ± 6.7 ms (YM); and low-frequency-to-high-frequency power ratio (LF/HF) was notably higher in YM (5.04 ± 3.52) than in elderly groups (OM: 1.53; OF: 1.08). Epoch accelerometer RMS was modestly higher in YM (0.317 g) than in elderly groups (OM: 0.306 g; OF: 0.302 g). The findings establish a quantitative framework for age-stratified noise modeling in wearable ECG devices, with direct implications for motion-robust R-peak detection algorithm design. The concurrent availability of ECG and inertial data from a single IEC 60601-certified device distinguishes this dataset from prior art and strengthens its value for signal processing research.