Effects of a Machine Learning-based Lower Limb Exercise Training System for Knee Pain

Development and Randomized Controlled Trial of an AI-powered Technological Surrogate Physiotherapist (TSP) Dedicated to Quality Enhancement and Cost Reduction in Knee Osteoarthritis Exercise Rehabilitation

The goal of the study is to confirm the idea of AI-powered Technological Surrogate Physiotherapist (TSP), by demonstrating its effectiveness and value as a new technology-based contribution to OA healthcare. Participants will be randomized to one of two groups: (1) the conventional PT group receiving the exercise program delivered through in-person sessions; or (2) the AI-guided group following the program through the TSP after an initial PT session. All individuals will take part in the study for 12 weeks, and data will be collected at baseline and 12 weeks after randomization.

Knee pain, often caused by osteoarthritis, is a prevalent musculoskeletal disorder among older adults and significantly reduces physical function and quality of life. Exercise therapy has been shown to be an effective form of treatment for knee pain. However, the traditional delivery of exercise therapy requires that individuals attend clinics to participate in face-to-face exercise sessions, which can be expensive and inconvenient. In recent years, information technologies have been used to support the delivery of exercise programs. The programs have also shown great benefits in improving the management of knee pain. However, it remains a concern that physical therapists are not able to provide the patients with direct and immediate supervision when exercises are taken place remotely at home or in community centers, which can be detrimental to exercise performance and the management of knee pain.

Thus, the research team has developed a machine learning-based exercise training system to provide evidence-based lower limb exercise videos, real-time movement feedback, and tracking of exercise progress for older adults with knee pain. In this study, a 12-week randomized controlled non-inferiority trial will be conducted to compare the effects of the AI-powered Technological Surrogate Physiotherapist with those of in-person physiotherapy sessions.

Inclusion criteria:

1. At least 50 years of age;
2. Report having knee pain most days of the week;
3. Has radiographic evidence of grade 2, 3, or 4 knee OA on the K-L scale in the posteroanterior and/or skyline view in at least 1 knee;
4. Willing and able, both physically and cognitively, to perform the exercises required in the study protocol;
5. Has normal or corrected-to-normal vision;
6. Able to speak Cantonese and read Chinese, AND
7. Able to provide written informed consent.

Exclusion criteria:

1. Having a history of knee or hip replacement surgery;
2. Being nonambulatory;
3. Having systemic inflammatory arthritis (e.g., gout);
4. Having a history of trauma (e.g., fractures around the knee, dislocation, sprains or tears of soft tissues like ligaments) or surgical arthroscopy of either knee within the past 6 months;
5. Having intra-articular injection to the knee within the past 6 months;
6. Having cognitive impairment;
7. Involvement in a similar study (i.e., involving physical exercise or physical therapies for knee pain) within the past 6 months;
8. Undergoing recent or imminent surgery (within 3 months) , OR
9. Having medical co-morbidities that preclude participation in exercise.