Terahertz metasurface biosensor for high-sensitivity salinity detection and data encoding with machine learning optimization based on random forest regression

This research presents a terahertz-based biosensor for high-precision salinity detection, employing a synergistic integration of graphene, gold, and silver in a metasurface configuration. The sensor exhibits exceptional performance characteristics within the 0.6–2 THz frequency range, demonstrating a maximum sensitivity of 571 GHzRIU⁻¹, quality factor of up to 6.107, and a figure of merit of 2.189 RIU⁻¹. Electromagnetic field distribution analysis demonstrates enhanced light-matter interactions, with peak performance observed at 1.6 THz. The investigation further explored the sensor's potential for multilevel data encoding, illustrating its versatility beyond conventional detection methods. Simulation results indicate the efficiency of the proposed sensor which surpasses some of the previous existing designs in terms of sensitivity among other performance parameters. A machine learning approach using random forest regression was employed to predict sensor responses, achieving coefficient of determination (R²) values ranging from 0.88 to 1.00 across multiple parametric studies. This research presents a promising platform for high-precision salinity sensing with potential applications in environmental monitoring, desalination processes, and aquaculture industries.