Robust and computationally efficient approach for Heart Rate monitoring using photoplethysmographic signals during intensive physical excercise

Abstract

Photoplethysmography (PPG) at the wrist can be used as a non-invasive technique for estimating the Heart Rate (HR). However, Motion Artifacts (MA) impede this estimation of the HR especially when the subject performs intensive physical exercise. This paper describes the implementation of a fast algorithm based on a statistical approach for extracting the HR from such MA corrupted PPG. Data-sets contain PPG signals with accelerometer data in 3-dimensions of subjects who were performing various forearm and upper arm activities. Simultaneously recorded Electrocardiogram (ECG) was used to verify the algorithm. The algorithm first de-noises the signal followed by a focus reduction in the spectrum and then peak tracking. The proposed algorithm resulted in an overall average-absolute-error of 1.5926 Beats Per Minute (BPM). The Bland-Altman was used to compare ECG derived HR and PPG derived HR where Limit of Agreement (LOA) was ±7 BPM. Also, this algorithm gives a great efficiency in time and robustness to track the HR on-the-go.