Plasmonic SPR biosensor with bayesian regression for non-invasive protein biomarker detection

This paper presents a designed metasurface sensor for protein biomarkers detection. The design of this metasurface structure is essentially dependent on the coating of four identical figure-eight-shaped resonators of MXene through black phosphorus and graphene. Meanwhile, the performance analysis demonstrates a competitive sensitivity (395 GHz/RIU) with a linear frequency response of R² = 0.954 and protein biomarker concentration (R² = 0.956). The sensor exhibits stable performance across 0.31-0.46 THz and maintains consistent operation at incident angles up to 30°. Interestingly, this tunability is achieved through graphene chemical potential modulation, with transmittance decreasing from 97.7% to 66.7% as its value increases. Additionally, the machine learning optimization using Bayesian Ridge Regression demonstrates exceptional predictive capabilities for both refractive index variations (R² ≈ 86%) and angular dependencies (R² ≈ 96%). This integrated photonic-microfluidic sensor platform offers significant potential for early detection of neurological disorders through rapid, sensitive, and scalable brain tumor biomarker analysis.