Abstract:
A computer-implemented method for estimating biophysiological rates using the Hilbert transform includes receiving a quasiperiodic data stream from a biophysiological sensor, and removing at least a portion of an offset from the quasiperiodic data stream to provide a smoothed data stream by filtering the quasiperiodic data stream through a bandpass filter and phase compensating the filtered quasiperiodic data stream. The method also includes transforming the smoothed data stream into an analytic data stream using a Hilbert transform approximation and calculating the time derivative of the phase angle of the analytic data stream, where the time derivative is a frequency of the quasiperiodic data stream. The method further includes providing an output data stream derived from the frequency.
Abstract:
A computer-implemented method for estimating biophysiological rates using the Hilbert transform includes receiving a quasiperiodic data stream from a biophysiological sensor, and removing at least a portion of an offset from the quasiperiodic data stream to provide a smoothed data stream by filtering the quasiperiodic data stream through a bandpass filter and phase compensating the filtered quasiperiodic data stream. The method also includes transforming the smoothed data stream into an analytic data stream using a Hilbert transform approximation and calculating the time derivative of the phase angle of the analytic data stream, where the time derivative is a frequency of the quasiperiodic data stream. The method further includes providing an output data stream derived from the frequency.
Abstract:
A computer-implemented method for estimating a cardiovascular feature, such as heart interbeat intervals, using time-delay embedding includes receiving a quasiperiodic observational data stream from a biophysiological sensor, selecting a first delay value, and generating a first vector with a first dimensionality in a multidimensional coordinate space from the observational data stream through time-delay embedding. The method also includes selecting a first plane with a second dimensionality equal to one less than the first dimensionality and which passes through an origin of the coordinate space, identifying a plurality of crossings of the first vector through the first plane in one direction, and correlating a time value corresponding to each of the crossings. The method further includes calculating a plurality of periods between the time values corresponding to consecutive crossings, where the periods represent a stream of interbeat intervals.