1435 - Video-Based Sensor Performance in a Real-World Rehabilitation Context: A Case Study

Objectives: The promise of “real-world” outcome measures is catalyzing rehabilitation research, and these measures are supported by a groundswell of advancing sensor technologies. Video-based sensor technology is one example that is gaining interest due to its potentiation of efficient recruitment and high-powered clinical research. However, understanding the limits of sensor technologies is important to research study design, to clinical translation, and to iterative improvement of the technology. For example, video-based sensor technology used in markerless motion capture is dependent on cameras and machine learning , which have respective limitations such as line-of-sight (cameras) and adequate training datasets (machine learning).

Design: In this case study, our objective is to explore performance of a video-based sensor technology for assessing upper limb biomechanics in a clinical setting. A single participant with chronic stroke was invited to an outpatient physiatry clinic for the video-based assessment. We apply repeated measures with this single participant for two conditions—namely, performing a reaching task with an applied orthosis and performing same reaching task without orthosis. Our primary outcome was feasibility of the video-based sensor technology to identify anatomic keypoints. As a secondary outcome, we compare biomechanical metrics obtained with orthosis and without orthosis present.

Results: Video-based sensor technology successfully identified anatomic keypoints of the participant for both orthosis conditions. Furthermore, data acquisition was well tolerated by the participant without adverse events.

Conclusions: This case study provides preliminary data regarding feasibility of video-based sensor technology to assess the upper limb of an individual with chronic stroke wearing a common orthotic intervention. Future prospective studies with diverse rehabilitation populations are needed to confirm feasibility of and to elucidate limitations of video-based sensor technology.