Assessing heart rate variability through wavelet-based statistical measures

Heart-rate variability (HRV), measured by the fluctuation of beat-to-beat intervals, has been growingly considered the most important hallmark of heart rate (HR) time series. HRV can be characterized by various statistical measures both in the time and frequency domains, or by nonlinear methods. During the past decades, an overwhelming amount of HRV data has been piled up in the research community, but the individual results are difficult to reconcile due to the different measuring conditions and the usually HR-dependent statistical HRV-parameters applied. Moreover, the precise HR-dependence of HRV parameters is not known. Using data gathered by a wearable sensor of combined heart-rate and actigraphy modalities, here, we introduce a novel descriptor of HRV, based on a modified Poincare plot of 24-h RR-recordings. We show that there exists a regressive biexponential HRV versus HR master curve (M-curve) that is highly conserved for a healthy individual on short and medium terms (on the hours to months scale, respectively). At the same time, we reveal how this curve is related to age in the case of healthy people, and establish alterations of the M-curves of heart-attack patients. A stochastic neuron model accounting for the observed phenomena is also elaborated, in order to facilitate physiological interpretation of HRV data. Our novel evaluation procedure applied on the time series of interbeat intervals allows the description of the HRV(HR) function with unprecedented precision. To utilize the full strength of the method, we suggest a 24-hour-long registration period under natural, daily-routine circumstances (i.e., no special measuring conditions are required). By establishing a patient's M-curve, it is possible to monitor the development of his/her status over an extended period of time. On these grounds, the new method is suggested to be used as a competent tool in future HRV analyses for both clinical and training applications, as well as for everyday health promotion.

Download Full-text

Effects of head-down bed rest on complex heart rate variability: response to LBNP testing

Journal of Applied Physiology ◽

10.1152/jappl.1994.77.6.2863 ◽

1994 ◽

Vol 77 (6) ◽

pp. 2863-2869 ◽

Cited By ~ 37

Author(s):

A. L. Goldberger ◽

J. E. Mietus ◽

D. R. Rigney ◽

M. L. Wood ◽

S. M. Fortney

Keyword(s):

Heart Rate ◽

Heart Rate Variability ◽

Frequency Domain ◽

Lower Body Negative Pressure ◽

Approximate Entropy ◽

Bed Rest ◽

Data Sets ◽

Adult Women ◽

Statistical Measures ◽

Physiological Variability

Head-down bed rest is used to model physiological changes during spaceflight. We postulated that bed rest would decrease the degree of complex physiological heart rate variability. We analyzed continuous heart rate data from digitized Holter recordings in eight healthy female volunteers (age 28–34 yr) who underwent a 13-day 6 degree head-down bed rest study with serial lower body negative pressure (LBNP) trials. Heart rate variability was measured on 4-min data sets using conventional time and frequency domain measures as well as with a new measure of signal “complexity” (approximate entropy). Data were obtained pre-bed rest (control), during bed rest (day 4 and day 9 or 11), and 2 days post-bed rest (recovery). Tolerance to LBNP was significantly (P < 0.02) reduced on both bed rest days vs. pre-bed rest. Heart rate variability was assessed at peak LBNP. Heart rate approximate entropy was significantly (P < 0.05) decreased at day 4 and day 9 or 11, returning toward normal during recovery. Heart rate standard deviation and the ratio of high- to low-power frequency did not change significantly. We conclude that short-term bed rest is associated with a decrease in the complex variability of heart rate during LBNP testing in healthy young adult women. Measurement of heart rate complexity, using a method derived from nonlinear dynamics (“chaos theory”), may provide a sensitive marker of this loss of physiological variability, complementing conventional time and frequency domain statistical measures.

Download Full-text