Terahertz metamaterial-based sensors have gained significant interest and emerged as one of the highly efficient sensing technologies in recent years. In particular, they have been proposed as an efficient tool for diagnosing cancer cells. Although numerous studies have demonstrated its potential for sensing and classification, this work, to the best of our knowledge, presents the first hexa-band metamaterial sensor for cancer cell detection using ML-based classification. In this study, a refractive index-based terahertz metamaterial sensor has been designed using a 5 \(\times\) 5 array concentric ring resonator structure for sensing blood cancer cell (n = 1.390), normal blood cell (n = 1.376), breast cancer cell (n = 1.399), normal breast cell (n = 1.385), skin cancer cell (n = 1.380), and normal skin cell (n = 1.360). The absorber exhibits six distinct peaks with absorption rates of 80, 93.5, 80, 65, 90, and 98% at 0.49, 1.38, 2.1, 2.9, 3.31, and 3.9THz, respectively. The calculated Q-factor (1.95, 6.25, 10.8, 22.07, 15, and 12.66), sensitivities (0.98, 0.64, 0.78, 0.56, 0.78, and 0.78 THz/RIU), and the figure of merit (FoM) (0.92, 3.04, 3.71, 4.3, 3.71, and 2.6) demonstrate improved sensing performance compared to previously reported meta-surface-based cancer cell sensors. A comprehensive dataset is generated by systematically varying the analyte thickness and refractive index to obtain diverse absorption spectra for cancer cell detection using machine learning. The classification abilities of different feature-based classifier models for cancer cells are compared using this dataset. Among these, the Extra Tree classifier performed with superior accuracy on both training and test data.
