Graphene-Enabled Multiresonator Metasurfaces for Ultrasensitive Surface Plasmon Resonance Detection of Waterborne Bacteria Across Multiple Frequencies with Machine Learning Optimization

This investigation presents the development and characterization of a high-performance terahertz biosensing platform optimized for ultrasensitive detection of waterborne pathogens across multiple spectral bands. The novel biosensor architecture integrates U-shaped and split circular strip resonators with advanced plasmonic materials, facilitating enhanced electromagnetic field confinement and interaction with analytes. Comprehensive computational analyses demonstrate exceptional sensing metrics, with the device achieving a maximum sensitivity of 565 GHzRIU⁻¹, a quality factor of 10.724, and a figure of merit of 3.55 RIU⁻¹ across three distinct frequency ranges of 0.15–0.2 THz, 0.57–0.65 THz, and 1.34–2 THz. To enhance sensor performance and predictive capabilities, machine learning methodologies were implemented, particularly utilizing decision tree regression algorithms. This computational approach demonstrates optimal coefficient of determination (R²) values of 100%, indicating exceptional predictive accuracy in transmission behavior. The biosensor demonstrates real-time, label-free detection capabilities with superior sensitivity compared to conventional water quality monitoring techniques. Furthermore, the platform’s versatility in detecting diverse waterborne pathogens and contaminants suggests significant potential applications in environmental monitoring systems and public health surveillance infrastructure.