RF-AE: Single-site Arterial Elasticity Estimation Using UWB Signals

Arterial elasticity is the capability of the arteries to dilate and constrict in response to fluctuations in blood pressure as the heart circulates blood throughout the body, serving as an important indicator for assessing arterial health. However, conventional medical detection methods are often complex, costly, and require direct contact between the user's skin and the device, making them impractical for regular monitoring. In this paper, we propose a contactless single-site arterial elasticity assessment approach using ultra-wideband (UWB) signals, where we estimate the brachial-ankle pulse wave velocity, a key indicator of arterial health, by analyzing the UWB signals reflected off the back of a subject sitting still. Specifically, we first propose preprocessing methods to reduce noise in the received UWB signals, and then design a deep generative adversarial network to generate pulse wave signals from these UWB signals. The generated signals are of high quality and fidelity, exhibiting characteristics of vasoconstriction and vasodilation. Further, we analyze the pulse wave and extract the features related to arterial elasticity from the generated pulse waves. Finally, we employ random forest regression to predict pulse transit time and a body-height based method to estimate the length of the pulse transit path, so as to achieve single-site arterial elasticity assessment. We build the corresponding system named RF-AE and conduct extensive experiments to evaluate its performance. The experimental results show that RF-AE can accurately predict the arterial elasticity of users.