Accuracy of cuffless blood pressure estimation using photoplethysmography and tonometry from pulse transit time alone

Background: Cuffless blood pressure (BP) estimation by pulse transit time (PTT) has advantages over standard cuff-based methods as it can be continuous and non-occlusive. However, the need for individual calibration between BP and PTT, and repeatability of calibrations are challenging.
Objectives: This pilot study investigates the repeatability and accuracy of BP estimation from PTT using contact-based photoplethysmography (PPG) and tonometry. Methods: Six subjects performed one-minute seated rest and 3-min isometric handgrip exercise on two consecutive days. Continuous finger BP was recorded. A PPG sensor was placed adjacent to a tonometer on the temporal pulse and a second PPG sensor on the left index finger and tonometer on left radial pulse. Beat-to-beat PTT was calculated for tonometry and PPG as the delay between the foot of temporal and wrist/finger signals. DBP/PTT linear regression on day 1 was used as calibration for prediction of DBP from PTT on day 2, and the regression from day 2 applied retrospectively for day 1. DBP was predicted using every 1, 5, 10, 20 and 30 beats observations. Predicted DBP was compared to measured finger DBP.
Results: Isometric exercise successfully increased DBP (average increase 23±11 mmHg). The minimum error of DBP estimation from PPG sensors was from 30-beats long observations (MAE±ESD=15±0.5 mmHg) and from tonometers was from 20-beats long observations (MAE±ESD=13±1 mmHg).
Conclusion: Cuffless BP estimated from PTT measured by PPG and tonometry provide a large error. In comparison, tonometry provides slightly better results as it requires less number of beats for estimation.