A Graph-based Method for Interbeat Interval and Heart Rate Variability Estimation Featuring Multi-channel PPG Signals During Intensive Activity

Inter-beat-interval (IBI) and heartrate variability (HRV) is important for numerous health monitoring applications. Although photoplethysmogram (PPG) sensors in wearables enable measurement of IBI, motion artifacts significantly impact the ability to accurately measure IBI. In this paper, we design a graph-based method to estimate IBI from motion-corrupted multi-channel PPG. We extract candidate heartbeats from noisy signals and leverage continuity in heartbeats to model them as a directed acyclic graph. IBI estimation is then modeled as a shortest-path problem in this graph. Our algorithm achieves percentage error of 4.33% and correlation of 0.94 for IBI estimation in motion-contaminated segments of PPG signals.