Pulse signal analysis based on wavelet packet transform and hidden Markov model estimation

The pulse signal can reflect the change of mechanisms and pathophysiology in the blood and viscera. An integrated approach, which combines the wavelet packet transform (WPT) with hidden Markov models (HMM), is presented to analyze the pulse signals, which often exhibit non-stationarity, in this study. Specifically, pulse signals measured from healthy and hypertensive subjects were decomposed into a number of frequency sub-bands, and energy features were then extracted from these sub-bands. The key features associated with each sub-band were selected based on the Fisher linear discriminant criterion. The key features were subsequently used as inputs to a HMM classifier for assessing the subjects' health status. Experimental results indicate that the proposed approach can differentiate the hypertensive pulses from healthy pulses effectively.