Comparison of Methods to Estimate Arterial Blood Pressure from the Pulse Arrival Time by Finger Photoplethysmography

Promising methods for monitoring blood pressure (BP) are based on the pulse arrival time (PAT) from finger photoplethysmography (PPG). This study aims to investigate the impact of PPG fiducial points, wavelengths, and PAT calibration models on short-term BP estimations. A multi wavelength multisensor PPG device was developed to measure PAT beat-by-beat in ten volunteers during an exercise protocol. Four calibration models (two-coefficient inverse, squared inverse, and logarithmic; and three-coefficient inverse), four fiducial points (wave onset, peak, maximum derivative, and tangents' intersection), and three wavelengths (red, infrared (IR), and green) were compared to estimate BP on 5-min segments and single beats. Finger systolic and diastolic BP (DBP) served as reference. Estimation errors were higher for systolic than DBP. Acceptable (<5 mmHg) median absolute errors (MAEs) for 5-min average systolic BP (SBP) were achieved by combining red or IR wavelengths with inverse or squared-inverse calibrations. For single-beat estimates, MAEs were higher but remained <8 mmHg with red or IR wavelengths, maximum derivative or tangents' intersection fiducial points, and two-coefficient calibrations. Short-term BP estimation from finger PAT is feasible by properly combining calibration models, wavelengths, and fiducial points. Identifying the optimal parameters for estimating BP from finger PAT by PPG, this study contributes to the development of noninvasive devices for accurate and continuous BP monitoring both in clinical settings and in the general population.