Kaia’s Computer Technology as Accurate as Physical Therapists

Kaia Health, a digital musculoskeletal company headquartered in Brooklyn, New York, has announced that a new clinical study published in the July 13, 2021, edition of the Journal of Medical Internet Research, has demonstrated that its computer vision technology is as accurate as a physical therapist (PT) in suggesting exercise corrections. The study is titled “App-Based Feedback for Rehabilitation Exercise Correction in Patients With Knee or Hip Osteoarthritis: Prospective Cohort Study.”

For this research, knee and hip osteoarthritis patients performed exercises while being simultaneously evaluated by computer vision technology and physical therapists. According to Kaia, “The results showed that physical therapists agreed as much with the exercise corrections suggested by Kaia Health computer vision technology as they did with each other’s exercise corrections.”

Kaia Health U.S. Chief Medical Officer Justin Yang, M.D. gave details to OTW on how the program gives automated, real-time exercise feedback by tracking exercise form through the user’s smartphone camera. “Our computer vision technology was developed specifically for the Kaia Health digital musculoskeletal app. It works using the front camera on a smartphone or tablet to capture data and process that data using machine learning algorithms.”

“Specifically, using a Pose Estimation Machine Learning Model that maps key points on the user’s body, Kaia Health computer vision technology can ‘understand’ what the user is doing (exercise pose) across multiple captured image frames in real time. Kaia Health’s machine learning model was trained using a proprietary image data set that consisted of data from people with a variety of characteristics exercising in front of their mobile devices, with a wide variety of exercise movements and environments. This allows the Kaia Health musculoskeletal app to provide immediate, personalized, audiovisual feedback on users’ exercise execution. As demonstrated within this study, the algorithm has proven to be highly reliable with both static and dynamic exercises across a range of motion and technical ability, resulting in technology that demonstrated to be as accurate as physical therapists in suggesting exercise corrections.”

Validated, Spot-On Accuracy

“Through feedback comparison between our Computer Vision tech and two in-person PTs from more than 500 exercises conducted by participants in an academic institution, results from this study validates our Computer Vision technology’s accuracy in providing audiovisual feedback to users are just as good as in-person PTs. Furthermore, Kaia’s technology is the first and only in this musculoskeletal industry to achieve this validation: our study results suggest that we are able to provide consistent and equal care for all, anywhere and anytime, no matter of age, gender, BMI [body mass index], pain location and exercise complexity. Together with our advanced AI [artificial intelligence] tech approach, Kaia is able to provide a frictionless, no sensor or additional hardware setup needed approach for users, bringing validated technology to the comfort of one’s home for musculoskeletal recovery.”

“With results from this publication, Kaia’s Computer Vision technology is now the only product in this MSK [musculoskeletal] market that provides a truly validated and evidence-based, convenient and scalable PT-grade access solution to our users.”

Noting a bit of surprise with some of the results, Dr. Yang told OTW, “We were expecting that we’d be getting this kind of results as this technology was developed by top-tier medical and computer vision experts and our exercise programs were rigorously evaluated by prior clinical trials. At the same time, I would say I’m surprised to see the consistency level to be this high and that the level of accuracy holds true for different scenarios, including age, gender, BMI, pain location and exercise complexity. This really suggests that our technology can provide equal care for all, anywhere, anytime.”