Heart Rate Monitoring During Intense Physical Activities Using a Motion Artifact Corrupted Signal Reconstruction Algorithm in Wearable Electrocardiogram Sensor

Accurate estimation of heart rates from electrocardiogram (ECG) signals during intense physical activity is a very challenging problem. In this study we investigated a novel technique to accurately reconstruct motion-corrupted ECG signals and HR based on time-varying spectral analysis. The algorithm is called Spectral filter algorithm for electrocardiogram Motion Artifacts and heart rate reconstruction (SegMA). The idea is to calculate time-frequency spectrum of ECG for each time shift of a windowed data segment and use the information from the spectrum to reconstruct HR during movement. The SegMA approach was applied to a datasets recorded in Chon Lab that includes 17 min recordings from 4 subjects during a challenging experimental protocol including walking, jogging, running, arm movement, wrist movement, body shaking, and weight lifting activities. The ECG and tri-axial accelerometer data were recorded from a wrist bands on both right and left wrists that are connected with wire through a tight suit. The reference ECG signals were recorded from chest using Holter monitor. The algorithm's accuracy was calculated by computing the mean absolute error between SegMA reconstructed HR from the wrist ECG and the reference HR from the Holter ECG. The average estimation errors using our method on this datasets are around 1 beats/min. These results show that the SegMA method has a potential for ECG-based HR monitoring in wearable devices for fitness tracking and health monitoring during intense physical activities.