Reneural

Neuro-rehabilitation patients play therapeutic Unity games (Ninja, Slingshot Wrist, Slingshot Pinch), and every session produces dense motion-capture data — 22 body joints at 15 Hz. But raw joint coordinates don't tell a physiotherapist what they need to know: which joints moved well, which side is weaker, and what should the next session focus on? Today that interpretation is manual, slow, and inconsistent — and most teams end up writing separate one-off scripts per game that can't be compared or reused. This project is a single, unified AI pipeline that converts any game's motion capture into a clinician-ready session report. It auto-detects the game from each file, cleans the data (dropouts, teleports, recording gaps), measures every joint movement in Python/NumPy — range of motion, speed, smoothness, repetitions, fatigue, left/right asymmetry — and then calls Meta Muse Spark to translate those metrics into focus areas with evidence, a next-session plan, and questions for the clinician. In short: it turns raw rehab-game recordings into an objective answer to the clinician's real question — what should we work on next?