Preliminary Finite Element Model for Hydrogen Peroxide-based Glucose Sensors.

Abstract

The development of continuous glucose monitoring and insulin control algorithms have enabled the recent development of closed-loop artificial pancreas technology. However, despite these advancements, glucose sensor accuracy and reliability under physiologic conditions and over long periods of monitoring continue to be limiting factors in achieving a truly closed-loop artificial pancreas. To develop improved sensor technology, glucose sensor dynamics and performance need to be modeled accurately under physiologic conditions. A three dimensional hydrogen-based glucose sensor model was developed to predict steady-state sensor performance. The finite element model, which included a three-electrode system and relevant electrochemical reactions for electrochemical current calculation, was developed on COMSOL Multiphysics software. The results were validated using an experimental setup measuring various hydrogen peroxide concentrations ranging from 5 mM to 35 mM. The model predicted a linear relationship between current ranging from $5 . 1 \mu A$ to $35 . 8 \mu A$ for the aforementioned glucose concentrations. Experimental data demonstrated a linear relationship between hydrogen peroxide concentration within the same range, and current measurements ranging from $9 . 4 \mu A$ to $60 . 6 \mu A$. The model and experimental data differed consistently by percentages between 40-46 % for all concentrationstested. This consistent scaling error can be attributed to the difference in electrode geometric area and electrochemical active area. Future iterations of the model must take into consideration the effective electrode area.