MICROST: A Mixed Approach for Heart Rate Monitoring During Intensive Physical Exercise Using Wrist-type PPG Signals

The performance of heart rate (HR) monitoring using wrist-type photoplethysmographic (PPG) signals is strongly influenced by motion artifacts (MAs), since the intensive physical exercises are common in real world. Few works focus on this study so far because of unsatisfying quality of corrupted PPG signals. In this paper, we propose an accurate and efficient strategy, named MICROST, which estimates heart rate based on a mixed approach. The MICROST framework is designed as a MIxed algorithm which consists of acceleration Classification (AC), fiRst-frame prOcessing and heuriStic Tracking. Experimental results using recordings from 12 subjects during fast running and intensive movement showed the average absolute error of heart rate estimation was 2.58 beat per minute (BPM), and the Pearson correlation between the estimates and the ground-truth of heart rate was 0.988. We discuss our approach in real time to face the applications of wearable devices such as smart-watches in reality.