Toward Continuous Estimation of Cardiorespiratory Parameters in Oscillometry: A Simulation Study

Abstract

Blood pressure's oscillometric waveform (OMW) comprises several cardiovascular components such as cardiac activity, respiration-related changes, and Mayer wave that contribute to its total variability over time. Accurate modeling of the OMW as a function of these components and continuous tracking of their underlying parameters can provide insights into the cardiovascular system dynamics and help determine the role played by each component in blood pressure variability. This paper presents a new state-space model for the OMW consisting of different parameters such as cardiac and respiration frequencies, amplitudes, and phases. Since the dynamic state-space model of the OMW is highly nonlinear and dependent on a large number of parameters, we utilized the extended Kalman filter (EKF). Since the EKF accuracy is highly dependent on the parameter's initial values, to obtain reasonable estimates of model initial values, a system identification procedure based on frequency domain analysis and curve-fitting was employed. The proposed method's performance was analyzed on simulated data with and without the proposed system identification procedure. A mean absolute percentage error of 2.68% was achieved in estimating OMW when using the proposed system identification approach. The proposed approach shows promise toward beat-to-beat tracking of cardiovascular parameters in oscillometric devices.