Arterial blood pressure parameter estimation and tracking using particle filters

In this paper, we present a computationally efficient method for adaptive tracking of physiological parameters such as heart rate and respiratory rate from the arterial blood pressure (ABP) measurement using particle filters. A previously reported estimation and tracking method was based on approximating the nonlinear models to linear ones based on the extended Kalman filters. However, the dynamic state-space model of the time-varying parameters and the ABP measurement is highly nonlinear in nature. In addition, the periodic nature of many of the time-varying parameters tend to make the estimation and tracking problem ill posed. In this light, the Rao-Blackwellized particle filtering method is proposed to adaptively estimate and track those parameters. The Rao-Blackwellized particle filter is capable of estimating the time-varying parameters of a nonlinear state-space model without performing any linear approximations while being computationally efficient. We demonstrate the performance improvements of our proposed method through computer simulations.