Design of Encoded and Tunable Graphene-Gold Metasurface-Based Surface Plasmon Resonance Sensors for Glucose Detection in the Terahertz Regime

This study presents the design, optimization, and evaluation of a highly sensitive terahertz refractive index sensor utilizing graphene metasurfaces for glucose detection in urine samples. The proposed sensor incorporates circular, square, and triangular resonator structures composed of graphene, gold, and silver to enhance plasmonic properties and sensing performance. Comprehensive parametric analysis and optimization were conducted using COMSOL Multiphysics simulations. The sensor demonstrates excellent performance characteristics, including a high sensitivity of 1000 GHzRIU-1 and quality factors ranging from 100.2 to 100.5. To further improve accuracy and reduce simulation time, an XGBoost Regressor model was integrated for predicting sensor behaviour across various parameters. The model achieved R2 scores consistently at or near 1, validating the robustness of the sensor design. Comparative analysis with existing literature highlights the superior sensitivity, figure of merit, and quality factor of the proposed sensor. This work contributes to advancing non-invasive glucose monitoring technologies and demonstrates the potential of machine learning integration in optimizing metamaterial-based sensors for biomedical applications.