Heart rate (HR) is one of the most commonly monitored physiological signals for patients in emergency rooms and intensive care wards. Average HR is physiologically significant on its own; however, quantities derived from the time-varying HR signal have proven more useful in certain clinical situations (ex. heart rate variability (HRV)). Unfortunately, the HR time series is poorly defined due to the randomly sampled nature of the signal, and current algorithms for estimating instantaneous HR are either non-physiological (contain stepwise discontinuities and ignore the cardiac refractory period) or inaccurate (spectrally smeared).
Given the need for a well-defined instantaneous HR signal, we developed a parametric model (integral pulse frequency modulation model driven by a continuous-time signal and physiological noise) and methods for estimating the model parameters from data. This poster describes the physiological basis for the model, the statistical framework upon which the model is built, and the algorithm for estimating model parameters.
The parametric model of instantaneous HR was developed and validated with simulations where the actual instantaneous HR signal was known. We show that instantaneous HR may be modeled using the statistical framework we developed and that model parameters can be extracted from available data with an acceptable error.