Scientists Can Now Detect Your Stress Just by Reading Your Brainwaves

NeuroStrata is a new EEG-based framework that analyzes mental stress by modeling how brain regions communicate over time, rather than relying on static features. Using 32-channel EEG from the SAM 40 dataset recorded during mental arithmetic tasks, the system builds time-varying connectivity maps and applies deep learning models to extract patterns. Beta-band connectivity proved most discriminating, hitting a peak accuracy of 97.3% with a Vision Transformer and support vector machine, while alpha-band connectivity stayed consistently stable across setups. The results highlight frontal-driven alpha influences and centrally integrated beta patterns tied to stress, with classification performance stabilizing in mid-to-late time windows—offering an interpretable, automated window into stress-related brain dynamics.