Electromagnetic Analysis and Machine Learning-Driven Optimization of a Graphene-Based Metasurface Sensor for Waterborne Pathogen Detection

Waterborne bacterial contamination represents a critical global health challenge, with traditional detection methods suffering from lengthy processing times, complex sample preparation, and limited real-time monitoring capabilities. This research develops a metasurface biosensor incorporating graphene-enhanced multilayer resonators for rapid, label-free detection of waterborne bacteria such as Escherichia coli, Vibrio cholerae, and Salmonella species. Electromagnetic analysis employing Maxwell’s equations, Kubo formalism for graphene conductivity, and coupled-mode theory provides comprehensive theoretical foundation for the sensor’s operation. Using COMSOL Multiphysics for numerical simulations, the sensor achieves a maximum sensitivity of 487.805 GHz/RIU and a figure of merit of 19.5122 RIU⁻¹ over the operating frequency range of 0.168–0.176 THz, demonstrating remarkable performance characteristics. As the sensitivity to changes in the refractive index increases, the detection limit drops substantially from 0.548421 RIU to 0.064244 RIU. Machine learning integration using polynomial regression algorithms achieves R² scores of 86% and 89% for predicting absorption characteristics based on graphene chemical potential and incident angle parameters. The proposed design establishes a scalable pathway toward real-time terahertz biosensing for next-generation water quality monitoring and microbial diagnostics.