Session SA2.4

A Point Process Approach to Assess Dynamic Baroreflex Gain

Z Chen*, EN Brown, R Barbieri

Massachusetts General Hospital
Boston, MA, USA

Evaluation of arterial baroreflex is an important topic in cardiology. A measure of baroreflex gain is essential in characterizing cardiovascular control and explaining both heartbeat dynamics and hemodynamics. Since the cardiovascular system has closed-loop interactions between many variables, research efforts have been devoted to estimating the baroreflex gain with a closed-loop system identification approach. However, most methods in the literature are batch-based with the assumption that the signals are stationary or locally stationary (within a moving window).
We present a point process approach to estimate the dynamic baroreflex gain within a closed-loop model of the cardiovascular system. Specifically, the inverse Gaussian probability distribution is used to model the heartbeat interval, whereas the instantaneous mean is modulated by a bivariate autoregressive model that contains the previous R-R intervals and systolic blood pressure (SBP) measurements. The dynamic baroreflex gain is estimated in the feedback loop with a point process filter, while the RR-to-SBP transfer function gain in the feedforward loop can be estimated by a Kalman filter. The adaptive filtering enables us to track the model parameters and to compute dynamic statistical indices including the HR, HRV, and baroreflex gain. The proposed two-filter estimation approach provides a quantitative assessment of interacting heartbeat dynamics and hemodynamics.
We apply our approach to real physiological recordings and evaluate the proposed model with goodness-of-fit tests. We first demonstrate our method using a dataset from the Physionet. The point process model passed the Kolmogorov-Smirnov test and provided a good fit of the data. Next, we apply the method to a tilt-table protocol study. We compare the estimated statistical indices between the "rest" and "tilt" conditions and conduct a rank-sum test. Results reveal statistical differences in these indices, including the baroreflex gain.
In summary, the proposed point process framework enables us to capture the transient dynamics of the HR and HRV and to model the dynamic nature of the baroreflex transfer function in a non-stationary environment. As a result, the statistical indices of HR, HRV, and baroreflex gain can provide a real-time noninvasive assessment for ambulatory monitoring in clinical practice.

(Abstract Control Number: 187)