An Efficient Method for Heart Rate Monitoring using Wrist-type Photoplethysmographic Signals During Intensive Physical Exercise

Abstract

Wrist type photoplethysmographic (PPG) signals are vulnerable to motion artifacts (MA), which affects the heart rate (HR) estimation. In this work, we have proposed an efficient method to estimate and track HR using PPG signals and simultaneous acceleration signals. In our method, an MA signal is generated by choosing one of the three axis acceleration signals based on their highest bandpower. An RLS filter is used to remove MA from the PPG signal, where the reference signal is the generated MA signal. A simple tracking and verification step is incorporated to give the correct value of HR in consecutive time windows. Implementing our proposed method on 12 subjects data during high speed running, we found the average absolute error of heart rate estimation was only about 1.23 beat per minute (BPM) with standard deviation of 1.92 BPM and the Pearson correlation coefficient between the estimates and the ground-truth of heart rate (obtained from simultaneously ECG) was .9956.

Publication
In 2016 5th International Conference on Informatics, Electronics and Vision (ICIEV), Dhaka, Bangladesh.
Date