Blood pressure estimation from photoplethysmography signals with and without calibration
Loading...
Author
Constantin, L.
Aguet, C.
Proença, M.
Bonnier, G.
Lemay, M.
Jorge, J.
DOI
Abstract
Blood pressure (BP) is a key cardiovascular biomarker, but its clinical value in acute settings depends on accurate, continuous monitoring. Recent approaches are exploring photoplethysmography (PPG) as a non-invasive BP monitoring solution, yet most of them require frequent recalibration using a cuff-based device - which curbs their practical utility - and are validated on small cohorts with limited BP variability. To address these limitations, we investigated both calibrated and calibration-free deep learning estimators using raw PPG and demographic data, tested on a large database with substantial BP variability. Our results highlight PPG as a robust option for tracking BP changes in contexts of high hemodynamic instability, while highlighting important differences between the two approaches.
Publication Reference
CSEM Scientific and Technical Report 2025, p. 17–18
Year
2025