ABSTRACT
Ultra-short-term (UST; <5 min) heart rate variability (HRV) is increasingly used to indirectly assess autonomic nervous system modulation and physical health. However, UST HRV estimates may vary with measurement technique, physiological state, and data preprocessing. The purpose of this investigation was to assess the information content of UST HRV and its sensitivity to different physiological states and preprocessing techniques. 26 time, frequency, and non-linear HRV measures were determined in 80 healthy men (age: 22.1 ± 3.7 yr) and 25 women (age: 19.4 ± 2.8 yr) from 2-min ECG recordings during seated and standing rest, low-intensity exercise, and seated recovery after maximal exercise. For men, HRV measures obtained during each condition were further analyzed with principal component analysis, k-means clustering, and one-way ANCOVAs. Backward stepwise regression was used to determine the ability of UST HRV to predict aerobic fitness. The sensitivity of UST HRV estimates to different artifact correction procedures was determined with intraclass correlation coefficients. Compared with men, women displayed HRV characteristics suggestive of greater vagal modulation. Nearly 80% of HRV information content was distilled into three principal components comprised of similar measures across conditions. K-means clusters varied in composition and HRV characteristics but not aerobic fitness, which was best predicted by HRV during standing rest. HRV estimates differed depending on artifact correction procedures but were generally similar after individualized correction. Our results indicate that UST HRV measures display redundancy but convey state-specific information and do not strongly predict aerobic fitness in healthy men. Most UST HRV measures are robust to slight differences in artifact correction procedures.
