Ravi Komatireddy, Anang Chokshi, Jeanna Basnett, Michael Casale, Daniel Goble and Tiffany Shubert
Introduction: Tele-rehabiliation technologies that track human motion could enable physical therapy in the home. To be effective, these systems need to collect critical metrics without PT supervision both in real time and in a store and forward capacity. The first step of this process is to determine if PTs (PTs) are able to accurately assess the quality and quantity of an exercise repetition captured by a tele-rehabilitation platform. The purpose of this pilot project was to determine the level of agreement of quality and quantity of an exercise delivered and assessed by the Virtual Exercise Rehabilitation Assistant (VERA), and seven PTs. Methods: Ten healthy subjects were instructed by a PT in how to perform four lower extremity exercises. Subjects then performed each exercises delivered by VERA which counted repetitions and quality. Seven PTs independently reviewed video of each subject’s session and assessed repetitions quality. The percent difference in total repetitions and analysis of the distribution of rating repetition quality was assessed between the VERA and PTs. Results: The VERA counted 426 repetitions across 10 subjects performing the four different exercises while the mean repetition count from the PT panel was 426.7 (SD = 0.8). The VERA underestimated the total repetitions performed by 0.16% (SD = 0.03%, 95% CI 0.12 - 0. 22). Chi square analysis across raters was χ2 = 63.17 (df = 6, p<.001), suggesting significant variance in at least one rater. Conclusion: The VERA count of repetitions was accurate in comparison to a seven member panel of PTs. For exercise quality the VERA was able to rate 426 exercise repetitions across 10 patients and four different exercises in a manner consistent with five out of seven experienced PTs.