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

Area of Science:

• Exercise Physiology
• Biomedical Engineering
• Cardiology

Background:

• Effective home-based exercise programs for older adults and individuals with chronic conditions need safeguards against exercise side effects.
• Exercise programs typically include rest, main exercise, and recovery phases.
• Automated methods for identifying these exercise modes have not been systematically investigated.

Purpose of the Study:

• To determine if time-domain heart rate variability (HRV) parameters differ across exercise modes (rest, main exercise, recovery).
• To identify the optimal time-domain HRV parameters for automated classification of exercise modes.
• To develop a classification model for automated exercise mode identification.

Main Methods:

• Discriminant analysis was employed to assess differences in time-domain HRV parameters between exercise modes.
• Selected HRV parameters were used to build a classification model.
• Classification success rate for identifying exercise modes was evaluated.

Main Results:

• Two specific time-domain HRV parameters, RRtri and MeanRR, were identified as significant differentiators of exercise modes.
• The identified HRV parameters yielded an 80% classification success rate in distinguishing between rest, main exercise, and recovery phases.
• Discriminant functions were generated for classifying exercise modes based on these HRV parameters.

Conclusions:

• Time-domain HRV parameters, specifically RRtri and MeanRR, can effectively differentiate between exercise modes.
• An automated classification model using these HRV parameters can achieve 80% accuracy in identifying exercise modes.
• This approach offers a potential safeguard for implementing home-based exercise programs, particularly for vulnerable populations.