Rewsyaydee/fyp-neurodrive-model — reverse-engineered prompt

Build me a desktop app called NeuroDrive that can monitor driving anxiety from EEG data, ideally from a Muse Athena headset in real time. I want it to connect to the live stream, extract the brain signal features it needs, run a trained machine learning model, and show a simple anxiety level or classification while the session is happening. Please make the interface easy to use, with a clean dashboard for live status, session info, and results.

I also want a way to train or retrain the model from saved data, then use the latest model in the app. After each session, save a report with the results and model performance details so I can review them later. If the headset is not available, include some kind of test or simulated mode so the app can still be demonstrated.

Please wire up whatever is already in this project, fix anything outdated or broken, and make the whole thing runnable end to end. You can check current docs online if needed.