Using heart rate variability for automated identification of exercise exertion levels

Heart rate monitoring is being used to estimate activity of autonomous nervous system by analysing heart rate variability (HRV). HRV has been recently shown to be effective means to monitor efficacy of exercise in patients with cardiovascular conditions and older adults. Whether HRV can be used to identify exercise exertion levels is unknown. There are multiple approaches to analyse HRV however it is not clear which approach is optimal in assessing cycling exercise. Previous studies demonstrated potential of analysis of short-term sequences of beat-by-beat heart rate data in a time domain for continuous monitoring of levels of physiological stress. The goal of this study was to assess the potential value of short-term HRV analysis during cycling exercise for automated identification of exercise exertion level. HRV indices were compared during rest, height of exercise exertion, and exercise recovery. Comparative analysis of HRV during cycling exercise demonstrated responsiveness of time-domain indices to different phases of an exercise program. Using discriminant analysis, canonical discriminant functions were built which correctly identified 100% of 'highest level of exertion' and 80.0% of 'rest' episodes. HRV demonstrated high potential in monitoring autonomic balance and exercise exertion during cycling exercise program.