<p>A simple, compact terahertz metamaterial absorber, analyzed with machine learning, is presented for cervical and adrenal cancer detection. The metamaterial absorber, fabricated with an aluminum layer on a polyimide substrate, exhibits a high sensitivity to the subtle variations in refractive index (<i>n</i>) between cervical cancer cells (<i>n</i> = 1.393), normal cervical cells (<i>n</i> = 1.368), adrenal gland cancer cells (<i>n</i> = 1.395), and normal adrenal gland cells (<i>n</i> = 1.381). The absorber demonstrates four distinct absorption peaks of 99.4% at 0.82597 THz, 89% at 1.152 THz, 95% at 1.516 THz, and 99.5% at 1.91 THz, all occurring at a unity refractive index. At resonant frequencies of 0.82597 THz, 1.152 THz, 1.516 THz, and 1.91 THz, the quality factors are 30, 8, 7, and 37 respectively. Additionally, the sensitivities across the four operating bands are 955 GHz/ refractive index unit (RIU), 183 GHz/RIU, 278 GHz/RIU, and 956 GHz/RIU respectively. Moreover, a machine learning model is developed using extra tree classifier to detect the different cancer cells. In this technique a huge dataset was generated corresponding to features like different incidence angle, polarization angle and thickness of the analyte. The model has an accuracy of 97.04% on validation data.</p>