AI-Augmented terahertz metamaterial biosensor for rapid and accurate isoquercitrin detection in herbal medicines

Abstract

We present a novel metamaterial-based terahertz biosensor integrated with AI for rapid isoquercitrin detection in herbal medicines. The sensor, optimized through COMSOL simulations, delivers exceptional sensitivity (300 GHz·RIU-1) and detects refractive index changes as small as 0.05 RIU. Its precision is validated by a near-perfect linear correlation (R2 = 99.73%) and stable performance metrics, including a 0.015 THz FWHM and Quality Factor of ∼47. Uniquely, a one-dimensional convolutional neural network augments predictive capability, achieving R2 values up to 1.00 across diverse conditions. This synergistic approach—combining terahertz spectroscopy, metamaterial-enhanced signal amplification, and AI-driven modeling—offers a transformative solution for standardizing and quality-controlling botanical therapeutics. By enabling fast, accurate, and scalable quantification of bioactive compounds, the system sets a new benchmark for analytical methodologies in natural product research.