A fully automatic algorithm for the analysis of heart rate changes and cardiac recovery during exercise

Exercise test is a challenging experimental condition for signal processing tools since stationarity requirement is not fulfilled. The aim of the present study was to develop a new way to look at HR and RR variability in exercise test, focusing both on exercise and the subsequent recovery phase. From RR series, the first derivative of the filtered tachogram and the variance were calculated in windows of predefined length and the three experimental conditions (rest, exercise and recovery) were automatically detected. Results indicate that in healthy subjects variance is larger during recovery than during exercise for a given RR interval, this is not seen in heart failure patients. In healthy subjects tonic and dynamical control of heart rate should be regarded as different autonomic regulations.