HEART RATE ESTIMATION USING WRIST-TYPE PHOTOPLETHYSMOGRAM SIGNALS CONTAMINATED BY MOTION ARTIFACTS USING ADAPTIVE FILTERS AND SPECTRAL PEAK TRACKING

• Main Author: Gani Ayub Tamudia, Put
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/36867
• Institution: Institut Teknologi Bandung

Abstract— Heart rate monitoring during physical activities can provide predictions about an individual’s health and physical performance. In clinical settings, this measurement may serve as a useful indicator for ambulatory patient monitoring. In sports, it can be used to generate personalized scheduling of training and recovery strategies. In military duties, real-time HR monitoring reveals body response in different hazardous situations, which extends to personnel alertness assessment and casualty detection. Real-time HR monitoring on wearable device platform often rely on chest strap electrocardiography (ECG) and wrist-type photoplethysmogram (PPG). When compared to ECG, wrist-type PPG provides a cheaper and more user-friendly alternative for wearable HR monitoring. However, photoplethysmogram signals are strongly affected by motion artifacts caused by body movements. This paper presents a method of estimating heart rate from wrist-type photoplethysmographic signals during fast running of increasing speed peaking at 15km/hour. The proposed method successfully suppresses the motion artifacts by using adaptive filters and Spectral Peak Tracking (SPT) algorithm. The proposed method produces HR estimations with mean absolute error of 1.17 beat per minute and the standard deviation of 0.95 beat per minute.