Blood Pressure Models for Wearable Sensors
preprintposted on 20.08.2021, 13:18 by Ana Carolina Gonçalves Seabra, Alexandre Ferreira da Silva, Thomas Stieglitz, Ana Belen Amado ReyAna Belen Amado Rey
As cardiovascular diseases are one of the most prominent illnesses, a continuous, non-invasive, and comfortable monitoring of blood pressure (BP) is indispensable. This paper investigates the best method for obtaining highly accurate BP values in non-invasive measurements when using an ultrasound (US) sensor projected for a wrist-worn device. State-of-the-art BP models were analyzed and qualitatively compared. Relevant arterial parameters such as luminal area, flow velocity and pulse wave velocity, of 729 subjects were extracted from a computer simulated database and served as input parameters for the wearable US. A linear in-silico model calibrated to each arterial-site revealed to be most accurate model. The linear model was used for the extraction of BP by using the US sensor and validated with a commercial pressure sensor in an ex-vivo experimental setup. The results showed an in-silico pulse pressure correlation of 0.978 and mean difference of (-2.134±2.477) mmHg at the radial artery and ex-vivo pressure correlation of 0.994 and mean difference of (0.554 ± 2.315) mmHg. Thus, with the linear model, the US measurement complies with the Association for the Advancement of Medical Instrumentation standard with deviations lower than 5 mmHg.