Continuous blood pressure monitoring from an autonomic nervous system perspective

Abstract

We propose a novel continuous blood pressure monitoring system which is based on an autonomic nervous system and which considers blood volume simultaneously since both affect blood pressure. An autonomic nervous system regulates blood pressure while blood volume is known to be proportional to the photoplethysmography (PPG) signal. To overcome the limitation of taking blood pressure using a conventional cuff inflating instrument, we designed a system which can achieve continuous blood pressure monitoring. In this research, we used a set of near-infrared light source (940nm) to create a divergent light which was collimated as a uniform beam incident to a wrist surface through a Fourier optics designed transfer lens. We found that the signals became more stable due to the uniform illumination and could be received by a detector. From the signals, we found that the blood volume when converted from blood velocity as measured by an ultrasound probe, showed a strong correlation with the signals. The heart rate variability analyzed from the signals, including time-domain (HR and SDNN) and frequency-domain (LF and HF) indices, could be viewed as physical models since these indices reflect the functions of an autonomic nervous system. Moreover, the research derived regression models can estimate blood pressure. Although it is not common to assess blood pressure from the perspective of an autonomic nervous system and blood flow simultaneously, our research approach seems logical. Our results show the potential for this novel system to be used for blood pressure health monitoring.

Integrating recurrent neural network (RNN) and Navier-Stokes equations for noncontact blood pressure assessment

“…Additionally, based on the pulse wave velocity (PWV), blood pressure values can be estimated [3]. Researchers have also used the Navier-Stokes equation to establish the relationship between blood flow and pressure, combined with heart rate variability indices to create regression models for mean arterial pressure [4]. Furthermore, many studies have leveraged machine learning or deep learning to build blood pressure models [5].…”