20 Sleep-Related Modulations of Heart Rate Variability…
317
techniques are otherwise applied in the analysis of weather data, water-level information, and stock-exchange prices, and are widely considered to be methods of fractal
analysis and chaos theory. For these approaches, the objectives are to analyze data
that appear coincidental and to detect an inner structure and patterns of order that
deviate from pure random behavior and that demonstrate phenomena of determinism.
One primary attempt here is to analyze the extent to which one value depends on
preceding values that happened seconds, minutes or even hours earlier. If there is
such non-random dependence among the values of a time series, the data are termed
to be auto-correlated.
One of the first applications of non-linear dynamics evaluated the complexity of
HRV throughout an entire night by analyzing beat-to-beat variability by means of
Wavelet- and Hilbert transform and found differences between healthy subjects and
patients with sleep apnea [17]. Subsequent studies on HRV applied detrended fluctuation analysis (DFA) to quantify short and long-term correlations but encountered
difficulties due to sudden jumps and fluctuations in the signals’ dynamics caused by
changes in sleepers’ positions at night, and during transition from one sleep stage to
the next. Such alterations render it impossible to find uniform patterns of behavior for
beat-to-beat variability in the heart rate. To enhance analysis, therefore, the course of
nocturnal heart rates was broken down according to the various sleep stages, and the
disturbances resulting from stage transitions were disregarded [6]. In other words,
sequences of pure sleep stages were prepared for study of heart rate. It was only in
the second step that the beat-to-beat sequences of heart rate were once again investigated for variability and sleep apnea. Investigations took place to quantify the extent
to which one heartbeat interval is correlated with subsequent heartbeats and revealed
distinct and highly pronounced differences between the sleep stages. Specifically, in
deep sleep, there is a virtually uncorrelated behavior pattern from heartbeat to heartbeat, whereas extensively correlated heartbeat behavior exists during REM sleep.
These differences were greater between the various sleep stages than the differences
found for episodes of heart rate with and without sleep apnea [6].
Initially, these results were surprising, since the influence of sleep apnea on heart
rate appears to be pronounced and distinct. However, the marked changes in sympathetic tone throughout the sleep stages provide a possible explanation. Changes in
sympathetic tone with respect to heart rate are not distinctly visible, since they are
smaller in amplitude. Still, they are highly apparent in the beat-to-beat variation
in heart frequency. This also explains why differences between the sleep stages as
determined by frequency analysis could in fact be determined. Indeed, in frequency
analysis, not only the frequencies are taken into account, but precisely also their
amplitudes (i.e., spectral power). Furthermore, our investigations of heart rate variability during sleep indicate that influences of autonomic tone on heart-rate regulation are so dominant that they still prevail during sleep apnea, and that they also
allow distinction to be drawn between sleep stages among these patients as well [6].
Therefore, with respect to beat-to-beat variability, the cyclical variation in heart rate
caused by sleep apnea merely signifies a relatively minor additional disturbance.
Accordingly, the results revealed the feasibility of a new procedure for determining,
from beat-to-beat regulation of heart rate, differences between sleep stages [28] with
317
techniques are otherwise applied in the analysis of weather data, water-level information, and stock-exchange prices, and are widely considered to be methods of fractal
analysis and chaos theory. For these approaches, the objectives are to analyze data
that appear coincidental and to detect an inner structure and patterns of order that
deviate from pure random behavior and that demonstrate phenomena of determinism.
One primary attempt here is to analyze the extent to which one value depends on
preceding values that happened seconds, minutes or even hours earlier. If there is
such non-random dependence among the values of a time series, the data are termed
to be auto-correlated.
One of the first applications of non-linear dynamics evaluated the complexity of
HRV throughout an entire night by analyzing beat-to-beat variability by means of
Wavelet- and Hilbert transform and found differences between healthy subjects and
patients with sleep apnea [17]. Subsequent studies on HRV applied detrended fluctuation analysis (DFA) to quantify short and long-term correlations but encountered
difficulties due to sudden jumps and fluctuations in the signals’ dynamics caused by
changes in sleepers’ positions at night, and during transition from one sleep stage to
the next. Such alterations render it impossible to find uniform patterns of behavior for
beat-to-beat variability in the heart rate. To enhance analysis, therefore, the course of
nocturnal heart rates was broken down according to the various sleep stages, and the
disturbances resulting from stage transitions were disregarded [6]. In other words,
sequences of pure sleep stages were prepared for study of heart rate. It was only in
the second step that the beat-to-beat sequences of heart rate were once again investigated for variability and sleep apnea. Investigations took place to quantify the extent
to which one heartbeat interval is correlated with subsequent heartbeats and revealed
distinct and highly pronounced differences between the sleep stages. Specifically, in
deep sleep, there is a virtually uncorrelated behavior pattern from heartbeat to heartbeat, whereas extensively correlated heartbeat behavior exists during REM sleep.
These differences were greater between the various sleep stages than the differences
found for episodes of heart rate with and without sleep apnea [6].
Initially, these results were surprising, since the influence of sleep apnea on heart
rate appears to be pronounced and distinct. However, the marked changes in sympathetic tone throughout the sleep stages provide a possible explanation. Changes in
sympathetic tone with respect to heart rate are not distinctly visible, since they are
smaller in amplitude. Still, they are highly apparent in the beat-to-beat variation
in heart frequency. This also explains why differences between the sleep stages as
determined by frequency analysis could in fact be determined. Indeed, in frequency
analysis, not only the frequencies are taken into account, but precisely also their
amplitudes (i.e., spectral power). Furthermore, our investigations of heart rate variability during sleep indicate that influences of autonomic tone on heart-rate regulation are so dominant that they still prevail during sleep apnea, and that they also
allow distinction to be drawn between sleep stages among these patients as well [6].
Therefore, with respect to beat-to-beat variability, the cyclical variation in heart rate
caused by sleep apnea merely signifies a relatively minor additional disturbance.
Accordingly, the results revealed the feasibility of a new procedure for determining,
from beat-to-beat regulation of heart rate, differences between sleep stages [28] with