UC Irvine

Stroke-induced neurological impairments affect millions globally, with movement-related disability one of the most common long-term consequences. Recent evidence suggests that proprioception – the ability to perceive the position and movement of body parts in space – plays a key role in movement rehabilitation after stroke. Proprioception is required for accurate control of the upper extremities during daily activities and also plays an important role as a teaching signal when learning new movement patterns. Stroke commonly impairs upper extremity proprioception, but there are few technologies available to help retrain proprioception. This thesis addresses this challenge by developing and testing a serious game – called “Proprio” – designed for at-home proprioceptive training using personal computer or smartphone technology, which offers a cost-effective, scalable solution.Proprio integrates real-time hand position tracking using the MediaPipe computer vision framework with an engaging, game-based exercise aimed at improving proprioception. The game is based on a popular, intuitive, and motivating music-based game (GuitarHero) in which users try to hit notes that stream across the screen. In Proprio, users hit notes by using the finger on one hand to point to a target finger or knuckle on the other hand at the correct moment indicated by streaming music notes. Computer vision detects the spatiotemporal accuracy of the pointing movements to drive game play. This thesis provides a detailed design description of Proprio and reports the results of an experimental study in which twelve individuals in the chronic phase of stroke trained with Proprio in the laboratory during two sessions spaced 1-2 days apart, followed by a follow-up assessment session one week later. Participants trained with a computer-based version of Proprio for 20–30 minutes per session. We evaluated improvements in proprioceptive-based pointing skill using gameplay performance and motion capture metrics. We also evaluated usability using a widely used measure, the System Usability Scale (SUS). Participants significantly increased their success at hitting notes over the three sessions by 12% +/- 5% SD and 7% +/- 7% SD when the target hand was the paretic hand or the unimpaired hand, respectively. Normalized pointing accuracy also improved significantly over time. Self-assessed technological proficiency predicted game performance (p < 0.05). The overall SUS score was 82.5 +/- 10 SD, indicative of a very satisfactory level of usability based on reference scores from a large database of other usability studies. These results demonstrate the feasibility and preliminary efficacy of using Proprio to train upper extremity proprioception. We have now created a version of Proprio that can be played on a smartphone. This research contributes to the growing field of digital health solutions for neurologic rehabilitation.