Are Pat-Based Models Suitable for Continuous Blood Pressure Monitoring in Ambulatory Settings?

Blood pressure (BP) is a critical biomarker for cardiovascular health. Several proposals of surrogate measurements and methods, such as Pulse Arrival Time (PAT), aim to provide non-invasive cuffless alternatives to traditional BP monitoring. However, the reliability of PAT-based models remains uncertain due to calibration challenges and the complex nature of BP regulation. This study investigated the performance of PAT-based models using portable electrocardiograms and photoplethysmography to estimate systolic (SBP) and diastolic (DBP) BP in ambulatory settings with 307 participants from the AuroraBP dataset. The best-performing PAT-based models achieved a mean absolute error (MAE) of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1 4. 2 3 ~ m m H g}$</tex> for SBP and 9.67 mmHg for DBP estimation. Both models failed to meet current international standards for BP monitoring. Additionally, the baseline models outperformed the best PAT-based model for SBP (MAE: <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1 3. 0 4 ~ m m H g}$</tex>) and achieved comparable performance to the top PAT-based model for DBP (MAE: 9.67). Despite promising results in previous studies, our findings highlight the poor performance of PAT-based models, particularly in dynamic and ambulatory conditions. These results highlight the limitation of multi-signal time-based models, which fail to generalize in complex dynamic situations. More advanced approaches that succeed in capturing the nonlinearities of BP variations and provide reliable, continuous BP monitoring in real-life settings are necessary.