Transfer Matrix–Based Design and Machine Learning–Assisted Optimization of a WS 2 /BaTiO 3 /Graphene/As 2 S 3 /ZnSe Multilayer SPR Biosensor for High‐Sensitivity Glucose Detection

Surface plasmon resonance (SPR) sensing is a well‐established technique for label‐free detection; however, the performance of conventional SPR platforms is often limited by inadequate electromagnetic‐field localization, optical dissipation, and restricted resonance tunability. To enhance sensing performance, a multilayer SPR structure composed of a BK7 prism, Ag film, WS 2 , BaTiO 3 , graphene, As 2 S 3 , and ZnSe layers is proposed and numerically investigated. The influence of the individual layer thicknesses on the optical response was systematically examined using the transfer matrix method (TMM) and evaluated through full‐wave electromagnetic simulations. The optimized configuration achieved numerically predicted absorption greater than 99% and exhibited strong field concentration at the sensing interface, indicating efficient plasmon excitation and enhanced light–matter interaction. The numerically obtained sensor response yielded a maximum local angular sensitivity of 500°/RIU, a figure of merit of 83.31 RIU −1 , and a detection limit of 0.015 RIU. In addition, machine learning models were developed to predict sensor responses for varying ZnSe thicknesses, graphene thicknesses, and analyte refractive indices. The predictive models demonstrated excellent agreement with simulation results, achieving coefficients of determination ( R 2 ) above 0.96 with low prediction errors. The obtained results confirm that the integration of WS 2 , BaTiO 3 , graphene, As 2 S 3 , and ZnSe effectively enhances the plasmonic response and sensing performance of the SPR platform, making it a promising design for high‐sensitivity refractive index sensing and advanced biosensing applications.