A differential autoregressive modeling approach within a point process framework for non-stationary heartbeat intervals analysis

Modeling heartbeat variability remains a challenging signal-processing goal in the presence of highly non-stationary cardiovascular control dynamics. We propose a novel differential autoregressive modeling approach within a point process probability framework for analyzing R-R interval and blood pressure variations. We apply the proposed model to both synthetic and experimental heartbeat intervals observed in time-varying conditions. The model is found to be extremely effective in tracking non-stationary heartbeat dynamics, as evidenced by the excellent goodness-of-fit performance. Results further demonstrate the ability of the method to appropriately quantify the non-stationary evolution of baroreflex sensitivity in changing physiological and pharmacological conditions.National Institutes of Health (U.S.) (Grant R01-HL084502)National Institutes of Health (U.S.) (Grant K25-NS05758)National Institutes of Health (U.S.) (Grant DP2-OD006454)National Institutes of Health (U.S.) (Grant DP1-OD003646)National Institutes of Health (U.S.) (Grant CRC UL1 RR025758