Integrating Dynamic Time Warping and K-means clustering for enhanced cardiovascular fitness assessment

Abstract

The appropriate measurement of cardiovascular ’Performance Enhancement’ is critical for improving physical fitness and overall health. Traditional methods, effective as they are, often rely on complex, cumbersome equipment, limiting their practical use for real-time, dynamic exercise evaluation. Thus, understanding the intricate relationship between physiological responses and exercise ‘Performance Enhancement’ using wearable technology is essential for tailoring effective fitness regimes. This study involved 52 participants, utilizing Garmin Vivosmart 5 wearables to analyze heart rate time series during the YMCA Three-Minute Step Test, assessing fitness levels and characterizing personalized heartbeat dynamics. The study employed Dynamic Time Warping (DTW) for clustering these time series into high and low VO2max groups. Additionally, heart rate dynamics were examined using K-means clustering to identify distinct patterns during exercise—namely ‘Efficient Adaptation’, ‘Balance Under Pressure’, ‘Active Strain’, and ‘Efficiency Improvement’ clusters. This analysis demonstrated that non-trained individuals showed higher ‘Active Strain’ and ‘Efficiency Improvement’ dynamics and lower ‘Efficient Adaptation’ dynamics, indicating the exercise’s varied effectiveness based on training level. This method provides a novel approach for identifying individual fitness levels and the efficacy of specific exercises, enabling personalized physical activity planning.