Estimating respiratory rate from instantaneous frequencies of long team heart rate tracings

Presents a method based on non-stationary state space modelling for the evaluation of the respiratory rate through the dynamical analysis of the instantaneous frequencies of heart-rate (HR). In practice, long term HR recordings are sequentially scanned by a sliding 20-sec window. Within each window, data are modelled using a singular value decomposition (SVD) based approach and the most important instantaneous frequencies, contained in the signals are estimated (the oscillation linked to the respiration is the 3rd lowest frequency). The presented method was tested by analysing 3 long term HR tracings (each lasting more than one hour) recorded in unrestrained subjects. Simultaneous direct recordings of the respiratory activity were performed and the data were used for comparing the algorithm performances. The obtained results indicate that the proposed algorithm is robust and efficient for the calculation of the instantaneous frequencies and may be a valuable tool for the estimation of the respiratory activity