Design and analysis of high-sensitivity hormone sensor with KNN behavior prediction for healthcare and biomedical applications

Abstract

This study presents an advanced sensor system integrating gold metasurfaces with graphene for the detection of reproductive hormones via refractive index variations. The proposed design operates under terahertz (THz) excitation to exploit the unique molecular signatures of hormones and minimize potential interferences. Comprehensive numerical analysis and finite element method (FEM) simulations were conducted to optimize the sensor’s design parameters and evaluate its efficacy. The optimized sensor exhibits exceptional performance metrics, including a peak sensitivity of 375 GHzRIU⁻1, a figure of merit (FOM) of 7.693 RIU⁻1, a quality factor (Q) of 2.381, and a limit of detection (LOD) of 0.556 RIU. Furthermore, the sensor demonstrates two-bit encoding capabilities through modulation of graphene’s chemical potential. Integration of a K-Nearest Neighbors (KNN) Regressor model enhances the sensor’s accuracy while reducing required resources and simulation time by approximately 85 %. These findings demonstrate the sensor’s potential for diverse applications, particularly in the biomedical field.