Enhanced Terahertz Graphene Metasurface Biosensor for Early Breast Cancer Detection with Machine Learning Optimization Based on Locally Weighted Linear Regression
摘要
This study presents surface plasmon resonance biosensor incorporating graphene and iron oxide (Fe₂O₃) metasurfaces for early breast cancer detection. The sensor design features a structured array of rectangular and square resonators optimized through comprehensive electromagnetic simulations. The sensor demonstrates a high sensitivity of 500 GHzRIU−1 across a refractive index range of 1.385–1.401 RIU, with figures of merit up to 6.494 RIU⁻1. The device maintains consistent spectral characteristics with a Full Width at Half Maximum of 0.077 THz and quality factors ranging from 6.571 to 6.610. Performance optimization using Locally Weighted Linear Regression (LWLR) achieved prediction accuracies of up to 100% R2. The sensor's enhanced sensitivity, coupled with machine learning optimization, presents a promising platform for non-invasive breast cancer detection.