Machine learning-assisted hybrid graphene–MXene–WS₂ terahertz metasurface biosensor for highly sensitive label-free chikungunya virus detection

Chikungunya virus (CHIKV) remains a major mosquito-borne disease requiring rapid and label-free diagnostic technologies for early detection This study presents a terahertz (THz) metasurface biosensor comprising tungsten disulfide (WS₂) square resonators, MXene concentric rings, gold (Au) circular resonators, copper (Cu) structures, and a gate-tunable monolayer graphene layer on a SiO₂ substrate for numerical investigation of refractive-index sensing using values representative of healthy and CHIKV-related plasma and platelet samples. Finite element simulations in COMSOL Multiphysics were performed to optimize the graphene chemical potential, incident angle, circular resonator radius, and square resonator dimensions. The optimized sensor exhibits a resonance frequency of 0.355 THz with strong electric-field confinement, achieving a maximum numerically predicted refractive-index sensitivity of 907 GHz/RIU over the representative refractive-index range of 1.33–1.39 RIU. The device maintains stable spectral performance for incident angles from 0° to 80°. XGBoost regression models accurately predicted the transmission response for varying structural parameters, achieving R 2 > 0.997 with low prediction errors. The present work represents a computational proof-of-concept demonstrating the feasibility of a THz metasurface for refractive-index sensing relevant to CHIKV detection. Experimental validation through device fabrication, biofunctionalization with CHIKV-specific recognition elements, and testing using clinical samples will be pursued in future studies to establish practical diagnostic performance.