Machine learning-enhanced terahertz plasmonic biosensor based on MXene-gold nanostructures for tuberculosis detection
Hussein A. Elsayed, Jacob Wekalao, Haifa A. Alqhtani, May Bin-Jumah, Mostafa R. Abukhadra, Stefano Bellucci, Amuthakkannan Rajakannu, Ahmed Mehaney
Abstract
This study presents a terahertz hybrid plasmonic biosensor utilizing MXene‑gold nanocomposites for tuberculosis detection. COMSOL Multiphysics simulations were employed to optimize sensor performance across varying chemical potential, incident angle, and resonator dimensions. The optimized configuration achieved a sensitivity of 1000 GHzRIU −1 and figure of merit of 22.22 RIU −1 , with a strong inverse linear relationship between resonance frequency and TB biomarker refractive indices (R 2 = 0.981). A machine learning framework based on decision tree regression was developed to predict sensor behavior, achieving R 2 values of 0.96, 0.92, and 0.88 for resonator dimensions, refractive index, and incident angle variations, respectively. The sensor platform offers significant potential for rapid, sensitive TB diagnostics in resource-limited settings.