A Novel Multimodal Physiological Model for the Noninvasive and Continuous Measurements of Arterial Blood Pressure

The wearable cuffless blood pressure (BP) technology is becoming an important tool for management and monitoring of hypertension which is the strongest risk factor for almost all different cardiovascular diseases (CVDs). Current cuffless BP estimation methods, however, mainly based on single or two modalities with the commonly used feature of pulse transit time (PTT), are inadequate to follow the BP variations and not accurate enough for clinical use. In this study, we propose a novel multimodal physiological model incorporating several parameters affecting hemodynamics such as skin temperature, cardiac cycle and arterial bioimpedance change/arterial diameter change, and PTT for arterial beat-to-beat BP estimation. These parameters can be easily extracted from photoplethysmographic, impedance-plethysmographic, electrocardiographic and temperature signals recorded simultaneously. This proposed mathematical model was validated on 23 human subjects during a cold pressor test which induced BP variations. The results of the experiment showed that the overall mean absolute error between the reference BP and the BP estimate with the proposed model were 5.78 and 4.15 mmHg for the systolic and diastolic blood pressures from all the subjects, respectively. The proposed multimodal model over-performed two highly cited PTT-based BP models with at least 32% improvement overall (statistically significant with p-value < 0.001). This work should facilitate the future development of medical-grade BP measuring devices for the precise diagnostics, management and monitoring of CVDs at home and hospital.