This paper introduces a tunable terahertz bandstop filter with an operating frequency of 2.11 THz. The filter design incorporates two symmetrically positioned split-ring resonators (SRRs) connected via a transmission line, achieving the desired bandstop characteristics. A graphene layer is employed within the structure to facilitate surface plasmon propagation and enable tunable filter performance. Simulation results validate the proposed filter’s capability to achieve the desired frequency response, with tunability over an 80 GHz range by varying the voltage applied across the graphene layer. Furthermore, the filter demonstrates exceptional sensitivity, achieving a maximum value of 0.18 THz/RIU, making it suitable for refractive index sensing applications. Machine learning-based regression models are utilized to predict the refractive index from the filter’s frequency response, serving as a reliable indicator of analyte property changes. The polynomial regression model outperformed other models, achieving a minimal mean square error of 0.00064. The proposed design showcases significant potential for advanced terahertz sensing and tunable filtering applications. Impact statement: The proposed tunable terahertz bandstop filter represents a significant advancement in the field of terahertz sensing and tunable filtering applications. By integrating graphene to enable surface plasmon propagation, the design achieves remarkable tunability across an 80 GHz range and exceptional sensitivity for refractive index sensing (0.18 THz/RIU). The use of machine learning regression models, particularly the polynomial regression model with a minimal mean square error of 0.00064, highlights the innovative approach to accurately predict analyte property changes. This work not only demonstrates a highly sensitive and tunable filter design but also bridges terahertz technology with machine learning, opening pathways for precise, real-time sensing solutions in fields such as biomedical diagnostics, chemical analysis, and environmental monitoring.