Symmetrical graphene–Au–Ag plasmonic terahertz metasurface with machine learning–assisted concentric resonators for high-sensitivity refractive index biosensing of colorectal and breast cancer biomarkers

• A hybrid graphene–gold–silver terahertz metasurface biosensor is proposed for label-free detection of colorectal and breast cancer biomarkers at picomolar–femtomolar sensitivity levels. • Strong electromagnetic field confinement is achieved using circular plasmonic resonators with concentric silver rings, resulting in a high refractive index sensitivity of 1000 GHz/RIU. • The biosensor demonstrates robust angular stability up to 80° incidence and tunable response via graphene chemical potential control (0.1–0.9 eV). • Distinct transmittance responses enable reliable discrimination between colorectal (1.329–1.401 RIU) and breast cancer biomarkers (1.385–1.401 RIU). • Machine learning–assisted regression confirms a strong linear resonance–refractive index relationship, validating sensing reliability and supporting non-invasive early cancer diagnostics aligned with SDG 3 (Good Health and Well-Being). Detecting cancer biomarkers at picomolar and femtomolar levels remains challenging due to the limited sensitivity of many clinical assays and their reliance on invasive sample collection. This work presents a terahertz metasurface biosensor integrated with machine learning for label-free identification of colorectal and breast cancer biomarkers. The proposed biosensor consists of a silicon dioxide substrate, a monolayer graphene layer, and hybrid plasmonic resonators formed using gold and silver. The metasurface comprises of circular resonators with a diameter of 500 nm arranged on a 10 μm graphene-coated square base, while concentric silver-coated rings with outer diameters of 1000 nm and 1500 nm to enhance electromagnetic field confinement in the terahertz regime. This hybrid graphene-plasmonic configuration increases light-matter interactions and improves the overall sensing performance. The graphene chemical potential is adjusted through full-wave electromagnetic simulations from 0.1 to 0.9 eV and the design sustains stable operation up to an incidence angle of 80°. A maximum refractive index sensitivity of 1000 GHz/RIU was obtained using geometric optimizations. Transmittance response differences were observed in the refractive index ranges 1.329–1.401 RIU for colorectal cancer biomarkers and 1.385–1.401 RIU for breast cancer biomarkers. The proposed design exhibits a Q factor > 1 and a figure of merit of 4.348 RIU −1 , signifying its appropriateness for low-volume and high-precision biosensing. The proposed graphene-based terahertz metasurface illustrates stable angular performance, high refractive-index sensitivity, and reliable discrimination of two clinically relevant cancer biomarker groups across a broad refractive-index window. The strong linear relationship between resonance frequency and refractive index established through both regression analysis and machine learning validation also validates the reliability of the sensing mechanism. By enabling non-invasive, early-stage cancer biomarker detection with minimal sample requirements, this work supports Sustainable Development Goal 3 (Good Health and Well-Being) as it provides a development of accessible and advanced diagnostic technologies for improved healthcare outcomes.