Glucose‐dependent dielectric Cole‐Cole models of rat blood plasma from 500 MHz to 40 GHz for millimeter‐wave glucose detection

Abstract In 2017, 30.3 million Americans lived with diabetes. Currently, the standard method of glucose estimation for diabetes relies on electrochemical sensors which have limited lifespans and are only reliable continuously up to 400 mg/dL. This paper investigates an alternative sensing method using dielectric property variability as a function of glucose concentration. In this study, we characterize dielectric properties—relative permittivity and conductivity—of Sprague Dawley rat blood plasma as they relate to plasma glucose concentrations from 250 to 16 000 mg/dL, over an expanded frequency range of 500 MHz to 40 GHz with a 9.897 MHz resolution. Particle swarm optimization is then used to fit measured data to a single‐pole Cole‐Cole model for relative permittivity and conductivity. Quadratic relationships are developed to represent Cole‐Cole parameters as functions of glucose concentration. Accuracy of the Cole‐Cole model is determined by finding the difference between experimental data and Cole‐Cole approximations. Results suggest three main conclusions which may be applied to future sensing applications and testing: the single‐pole Cole‐Cole model is a reliable method to accurately describe and reconstitute dielectric properties according rat blood plasma glucose concentrations, conductivity is a more promising determinant of glucose concentration than relative permittivity, and higher frequency ranges result in larger conductivity contrast between glucose concentrations.