Lol that's actually what we used initially (still do). We rolled our own way of processing the EEG signals (which are notoriously noisy), then we used a technique to remove movement.

Now typically those type of analysis are done post run. However,I worked on it for a while and now in real-time we can do error correction based on movement. This enables those EEGs to be within spitting distance of medical grade EEGs.