Early and accurate detection of breast tissue abnormalities is essential for improving breast cancer prognosis. This work presents a terahertz (THz) biosensor based on a hybrid graphene–vanadium dioxide (VO \(_2\) ) architecture designed to achieve high refractive-index sensitivity in the presence of healthy and malignant breast tissues. The electromagnetic response of the structure is modeled using the Wave Concept Iterative Process (WCIP), enabling a rigorous characterization of resonance shifts induced by the dielectric properties of fat, fibrous, and tumor tissues. The proposed WCIP implementation shows close agreement with analytical predictions, confirming the accuracy and efficiency of the modeling approach. The biosensor exhibits distinct resonance shifts for the different tissues, resulting in sensitivities of 0.81 THz/RIU, 0.84 THz/RIU, and 0.92 THz/RIU for healthy fibrous, healthy fat, and tumor tissues, respectively, along with competitive figures of merit and quality factors. To further investigate the impact of material parameters on performance, a deep neural network (DNN) model is trained on WCIP-generated data to predict sensitivity from graphene chemical potential, relaxation time, temperature, and VO \(_2\) conductivity. The DNN achieves high predictive accuracy, with \(R^2\) values above 0.97 for all tissue types under multiple train–test configurations, demonstrating its ability to capture nonlinear parameter interactions. The combination of high sensitivity, compact geometry, and AI-driven performance prediction highlights the potential of the proposed sensor as a promising platform for non-invasive breast cancer detection in the THz spectrum.
