Classification of cycling exercise status using short-term heart rate variability

Introduction of effective home-based exercise programs in older adults and people with chronic conditions requires implementation of appropriate safeguards to prevent possible side effects of strenuous exercise. In each exercise program the following exercise modes can be generally recognized: rest, main exercise, and exercise recovery. However, approaches for automated identification of these exercise modes have not been studied systematically. The primary purpose of this study was (1) to assess whether time-domain HRV parameters differ depending on exercise mode; (2) to identify optimal set of time-domain parameters for automated classification of exercise mode and build a classification model. Using discriminant analysis, two HRV parameters (RRtri and MeanRR) were identified which yielded 80% classification success in identifying correct exercise mode by applying generated discriminant functions.