A Fused Learning and Enhancing Method for Accurate and Noninvasive Hydration Status Monitoring With UWB Microwave Based on Phantom

Abstract

Hydration status is a critical indicator for human health, yet there are relatively few studies on monitoring the hydration status of the human body. Traditional methods for hydration monitoring have shortcomings such as being invasive, time-consuming to detect, and not being real time. Recently, microwave hydration monitoring has attracted research interest widely. This approach is noninvasive and based on the change of dielectric properties of human tissues with different hydration statuses. In this article, a fused learning and enhancing method is proposed to accurately monitor the hydration by using an ultrawideband (UWB) microwave. In the proposed method, an Elman neural network learning model and a joint enhancing method are designed. The joint enhancing algorithm is composed of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$k$</tex-math> </inline-formula> -medoids and least squares. To verify the performance of the proposed method, a muscle phantom is developed using bovine serum albumin. By changing the concentration, the phantom can mimic the muscle with different hydration statuses. In the evaluation experiment, the S-parameter and frequency data were measured with UWB microwave. Then, the data were employed for assessing the level of hydration with the proposed method. Experimental results showed that the relative error for estimating the concentration of the muscle phantom is less than 3.3%. These results demonstrated the feasibility of the proposed hydration monitoring method and proved that it is promising for monitoring human hydration status.