External Measurement of Swallowed Volume During Exercise Enabled by Stretchable Derivatives of PEDOT:PSS, Graphene, Metallic Nanoparticles, and Machine Learning

Abstract Epidermal sensors for remote healthcare and performance monitoring require the ability to operate under the effects of bodily motion, heat, and perspiration. Here, the use of purpose‐synthesized polymer‐based dry electrodes and graphene‐based strain gauges to obtain measurements of swallowe...

Full description

Bibliographic Details
Published in:Advanced Sensor Research
Main Authors: Beril Polat, Tarek Rafeedi, Laura Becerra, Alexander X. Chen, Kuanjung Chiang, Vineel Kaipu, Rachel Blau, Patrick P. Mercier, Chung‐Kuan Cheng, Darren J. Lipomi
Format: Article
Language:English
Published: Wiley-VCH 2023-04-01
Subjects:
Online Access:https://doi.org/10.1002/adsr.202200060
Description
Summary:Abstract Epidermal sensors for remote healthcare and performance monitoring require the ability to operate under the effects of bodily motion, heat, and perspiration. Here, the use of purpose‐synthesized polymer‐based dry electrodes and graphene‐based strain gauges to obtain measurements of swallowed volume under typical conditions of exercise is evaluated. The electrodes, composed of the common conductive polymer poly(3,4 ethylenedioxythiophene) (PEDOT) electrostatically bound to poly(styrenesulfonate)‐b‐poly(poly(ethylene glycol) methyl ether acrylate) (PSS‐b‐PPEGMEA), collect surface electromyography (sEMG) signals on the submental muscle group, under the chin. Simultaneously, the deformation of the surface of the skin is measured using strain gauges comprising single‐layer graphene supporting subcontinuous coverage of gold and a highly plasticized composite containing PEDOT:PSS. Together, these materials permit high stretchability, high resolution, and resistance to sweat. A custom printed circuit board (PCB) allows this multicomponent system to acquire strain and sEMG data wirelessly. This sensor platform is tested on the swallowing activity of a cohort of 10 subjects while walking or cycling on a stationary bike. Using a machine learning (ML) model, it is possible to predict swallowed volume with absolute errors of 36% for walking and 43% for cycling.
ISSN:2751-1219