Personalized Cuffless Blood Pressure Measurement based on Arterial 0D-1D Model with Carotid Flow and Radial Pulse

Cuffless blood pressure (BP) shows great potential to revolutionize the management of hypertension-one of the most modifiable risk factors of cardiovascular diseases. Extensive studies take advantage of peripheral photoplethysmogram (PPG) signal for cuffless continuous BP measurement, and demonstrated reasonable performance. However, the PPG signal is usually acquired from a single locality and it only reveals the information of blood volume change, resulting in poor generalization across different individuals for BP estimation. In this study, we propose a coupled arterial 0D-1D model for brachial BP estimation, incorporating carotid blood flow and radial artery PPG signals that represent both central and peripheral cardiovascular information for personalized modeling. We validated the proposed model on the Pulse Wave Dataset data that include 4374 samples. The results show that the mean error ± standard deviation for the estimated systolic and diastolic BP were 1.39 ± 2.53 mmHg and -0.47 ± 2.39 mmHg, respectively. Comparison with machine learning models such as decision tree and random forest also demonstrated that the 0D-1D BP estimation model could integrate multimodal signals and combine with blood flow path information to make personalized BP measurement.