Time delay estimation between two biosignals using system identification

Pulse Transit Time (PTT) is a marker of arterial stiffness and may allow continuous, non-invasive, and cuff-less monitoring of blood pressure. In this work, robust PTT estimation is sought by application of system identification to arterial and toe photoplethysmogram (PPG) waveforms. To demonstrate the feasibility of the concept, an auto-regressive with exogenous input (ARX) model was selected. The system identification procedure was applied to clean PPG signals collected from seven healthy subjects in the sitted position. The peak-to-peak detection algorithm was used for comparison. Our results show that using system identification, the coefficient of variation for two from seven of the subjects signals are much lower (about 0.05 for system identification and 0.35 for peak-to-peak) compared to the peak-to-peak algorithm.