Exploring the Wearable and Embeddable Solutions for Biopotential Signal Measurement: Dry and Non-Contact Technologies

With a rising interest in wearable solutions and smart fitness gadgets, unobtrusive monitoring of physiological parameters is a topic gaining more and more interest. Being applied over clothes or embedded in the fabric, non-contact capacitive electrodes present a particularly interesting solution. However, in the context of digitalization of the healthcare industry, the question of the possibility of combining such a biomedical signal acquisition with artificial intelligence for postprocessing is posed. Instead of using biomedical signals generators, in this paper, capacitive electrodes with high input impedance are designed and used for real-life acquisition of electrocardiograms (ECGs), measured on thumbs over a layer of thin cotton gloves. These measurements are then used to analyze the influence of geometrical and fabric parameters on the quality of measured biomedical signals. For this purpose, instead of noise analysis, a novel approach was adopted: descriptive statistical features are used to evaluate the performance using four machine learning classifiers. The results have confirmed the possibility of classification and paved the way to further investigation of employing machine learning for capacitive biomonitoring. Furthermore, it is confirmed that tighter coupling, thicker fabric of lower dielectric permittivity, and larger detection surface area ensure better performance.