Heart rate variability : characterizing autonomic control

Abstract

The overall goal of this research was to establish whether parameters derived from heart rate (HR) variability could be utilized to characterize autonomic states. Specifically, to (1) identify those parameters, derived from R-R interval data, which could be used as measures of dynamic and static autonomic tone, (2) determine the shortest signal duration required to achieve a valid measure of autonomic status, (3) utilize the information from the first two objectives in correctly classifying various autonomic states, (4) compare pharmacologically induced autonomic states with naturally occurring autonomic responses, and (5) characterize the daily changes resulting from evoked stressors. One study involved pharmacological agents which were administered to dogs in a five day study designed to establish specific dynamic and static autonomic conditions. Thirty-two parameters, derived from the frequency spectrum of HR variability waveforms calculated from electrocardiographic signals, were statistically analyzed to evaluate each parameter's ability to classify the three dynamic and three static autonomic conditions. A 94% accuracy in classifying these states was established using cluster analysis of two parameters: HR and power distribution of the low frequency (0.03125 to 0.0625 Hz) components of the HR variability waveform. Thirty-two seconds of HR variability data were needed for this classification. A second study of five days evaluated the day to day autonomic responses resulting from a stressor (air horn) and the anticipatory stress existing immediately before the treatment. Autonomic responses did result from the stressor, but day to day changes were difficult to characterize; anticipatory stress increased for two days, following the stressor, and then decreased on the last day. One of the parameters which proved to be a good discriminator in the pharmacological studies, power distribution of low frequency HR variability components, was significant in these studies. However, sixty-four seconds of data were needed for optimal discrimination...