Advanced terahertz metamaterial sensor for biomedical diagnostics featuring multi-material coupling, graphene control, and predictive AI models

Terahertz (THz) metamaterial biosensors have emerged as promising platforms for non-invasive and label-free biomedical diagnostics due to their strong electromagnetic field confinement and spectral sensitivity. However, existing designs often suffer from limited tunability, moderate quality factors, and insufficient sensitivity for detecting low-concentration biomarkers. In this work, a high-performance multi-material terahertz metasurface sensor is proposed, operating within the 0.1–0.75 THz frequency range for advanced biomedical sensing applications. The proposed architecture integrates graphene with silver, gold, and carbon nanotube (CNT)-based resonators arranged in a quadrantal symmetric configuration, enabling strong electromagnetic coupling and enhanced field localization. Dynamic tunability is achieved by varying the graphene chemical potential from 0.1 to 0.9 eV. Numerical simulations performed using COMSOL Multiphysics demonstrate stable operation up to an incident angle of 80°, with improved impedance matching and quality factors exceeding 3.0. The sensor exhibits high refractive index sensitivities of up to 600 GHz/RIU for malaria biomarker detection and 200 GHz/RIU for alcohol sensing, along with a minimum detection limit of approximately 0.312 RIU. Furthermore, an XGBoost-based machine learning model is employed to predict sensor responses under varying refractive index, chemical potential, and angular conditions, achieving prediction accuracies exceeding 94%. The proposed design demonstrates strong potential for point-of-care diagnostics and early disease detection, supporting the development of intelligent, high-sensitivity terahertz biosensing systems aligned with sustainable healthcare objectives.