Method, computer program product, computer-readable storage medium and system

A system for processing a signal representing a physiological rhythm of a subject, the physiological rhythm having a period characteristic (T, ti), the system comprising:

• a sensor disposed on the subject, where in the sensor is an accelerometer that generates the signal comprising raw three-dimensional acceleration data;
• and
• a computer, wherein the computer is in communication with the sensor, wherein the computer is configured to;
• receive the raw three-dimensional acceleration data from the sensor;
• preprocess the raw three-dimensional acceleration data by summing axes to form a one-dimensional signal and bandpass filtering the one-dimensional signal to define the signal;
• estimate a first group of period characteristic (T, ti) estimations based on the signal;
• generate a prior probability (P(θ)) for the period characteristic (T, ti) based on at least a subset of the first group of period characteristic (T, ti) estimations;
• estimate a second group of period characteristic (T, ti) estimations based on the first group and the prior probability (P(θ));
• determine a spectral likelihood (fS) of at least part of the signal, wherein the spectral likelihood (fS) is representative of a likelihood that the first group of period characteristic (T, ti) estimations corresponding to the at least part of the signal represents the period characteristic (T, ti) of the physiological rhythm, wherein estimating the second group of period characteristic (T, ti) estimations is based on the spectral likelihood (fS) of the signal;
• define a localization point in the physiological rhythm, wherein the localization point characterizes a phase in a period of the physiological rhythm;
• determine a temporal likelihood (fT) of the signal corresponding to the period, wherein the temporal likelihood (fT) is representative of a likelihood that the signal represents the localization point;
• and
• estimate the localization point based on the spectral likelihood (fS), the temporal likelihood (fT), and the prior probability (P(θ)).