Surface Plasmon Resonance (SPR) and Localized Surface Plasmon Resonance (LSPR) are fundamental mechanisms for high-performance optical sensing for diabetes screening and early detection. However, conventional planar SPR sensors often exhibit limited sensitivity for nanoscale and localized biomolecular interactions. This study presents a novel WS₂–graphene metasurface biosensor validated numerically for improved glucose detection in the terahertz (THz) regime for quantitative glucose monitoring in diabetes management. The proposed sensor consists of a multilayer architecture incorporating micro-/nanoscale resonators with square, rectangular, and circular geometries conformally coated with WS₂ and graphene on a glass substrate. This hybrid configuration allows simultaneous excitation of SPR and LSPR modes, providing enhanced electromagnetic field confinement and improved refractive index sensitivity. Using COMSOL Multiphysics, the electromagnetic response of the sensor is systematically validated by varying the graphene chemical potential (0.1–0.9 eV), incidence angle (0°–80°), and resonator geometry. The optimized metasurface provides excellent performance for glucose sensing within a refractive index range of 1.335–1.347, corresponding to different glucose concentrations relevant to diabetes screening and detection. The sensor achieves a maximum sensitivity of 2000 GHz/RIU, a stable full width at half maximum (FWHM) of 0.391 THz, and a figure of merit (FOM) of 5.115 RIU⁻¹. Furthermore, machine learning (ML) models trained on refractive index and incidence-angle variations exhibit near-perfect predictive capability with R² > 0.9997 and minimal prediction errors (RMSE ≈ 0.0014–0.0020), validating the stability and predictability of the sensor’s electromagnetic response. Comparative analysis with previously reported graphene-based and metal-composite sensors confirms the superior sensing performance of the proposed WS₂–graphene metasurface. The developed platform offers a promising approach for precise biochemical sensing applications, with strong potential for future real-time non-invasive glucose screening, early diabetes detection, and continuous diabetes management.
K et al. (Mon,) studied this question.