Key result
Novel PPG and ECG residual network outperforms existing models with ~4 mmHg SBP estimation error.
Why the study?
Demand for cuffless blood pressure monitoring is increasing, with research focused on enhancing estimation accuracy using intelligent algorithms to estimate systolic and diastolic blood pressure from photoplethysmogram and electrocardiogram signals.
Does a novel residual network framework using PPG and ECG signals improve the accuracy of cuffless blood pressure estimation compared to existing models?
Does a novel residual network framework using PPG and ECG signals improve the accuracy of cuffless blood pressure estimation compared to existing models?
A novel deep learning framework using PPG and ECG signals provides highly accurate cuffless blood pressure estimation that meets international standards.
May facilitate cuffless blood pressure monitoring; leaves open clinical adoption pending prospective validation.
The demand for cuffless blood pressure monitoring is increasing in the clinical management of cardiovascular events. Currently, cuffless non-invasive devices for blood pressure control are widely used, with the primary research focus on enhancing the accuracy of blood pressure estimation using intelligent algorithms. This study aims to explore a cuffless method for estimating systolic blood pressure (SBP) and diastolic blood pressure (DBP) using photoplethysmogram (PPG) and electrocardiogram (ECG) signals. To this end, we propose a novel residual network framework based on dynamic convolution strategy and multi-scale convolutional attention enhancement mechanisms. The framework first utilizes dynamic convolution and channel adaptive mechanisms to extract low-level features from the PPG and ECG modalities. It then employs a novel parallel cross-mixing architecture that integrates residual blocks with a multi-scale attention block (MSAB), which complements a global attention mechanism to enhance the modeling of long-range dependencies. The proposed framework was validated on the VitalDB dataset, and experimental results demonstrate excellent performance in both SBP and DBP estimation tasks (SBP MAE ± SD: 3.83 ± 5.71 mmHg, DBP MAE ± SD: 2.29 ± 3.69 mmHg). The framework outperforms existing models and meets international blood pressure measurement standards, including those set by the American Association for Medical Instrumentation (AAMI).
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2025) studied Blood pressure estimation. Multi-scale convolutional attention enhanced residual network vs. Existing models was evaluated on Systolic and diastolic blood pressure estimation accuracy (MAE ± SD). A novel residual network framework using PPG and ECG signals achieved excellent blood pressure estimation accuracy (SBP MAE 3.83 ± 5.71 mmHg, DBP MAE 2.29 ± 3.69 mmHg), outperforming existing models.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: