On the Design and Implementation of a Highly Accurate Pulse Predictor for Exercise Equipment

Goal: This study aims to develop highly accurate heart rate monitoring from the hand-held contact signal within a noisy environment during exercise. Methods: The periodic pattern and uncertainties of a physiological signal are modeled by a Laplacian random process. Based on this statistical model, a highly accurate pulse predictor (HAPPEE) is derived and implemented in real-time on a Cypress PSoC 5LP development board. A real-time experiment is designed to compare HAPPEE with a commercial heart rate monitor from POLAR. The percentage of credible estimates and the mean square error (MSE) of credible estimates are reported for experiments with seven healthy subjects. Results: The overall percentage of credible estimates is 99.2% for HAPPEE and 93.6% for POLAR. The overall MSE of credible estimates is 3.1 for HAPPEE and 7.7 for POLAR. These results show that HAPPEE is more accurate than POLAR. Conclusion: HAPPEE is able to accurately monitor heart rate within a noisy environment during exercise. Significance: Unlike existing heart rate estimation methods, HAPPEE does not require pulse detection or tuning parameters. It can be easily implemented in real-time on a low power and low cost development board for exercise equipment and outperforms a commercial heart rate monitor.