Knee4Life Project: Empowering Knee Recovery After Total Knee Replacement Through Digital Health

1. Background Total knee replacement (TKR) is a common procedure, with more than 100,000 per year undertaken in England and Wales. Although most patients have a successful outcome following their TKR, approximately 10-20% of patients are dissatisfied, chiefly because of pain and knee stiffness. A method to detect early problems with pain and stiffness could facilitate earlier referral to non-surgical treatments, which are effective in preventing the need for manipulation under anaesthetic [MUA]. Currently, rates of MUA are 2.5% (~2,500 patients per year in England and Wales), costing ~£14k per procedure. Our current understanding of when stiffness develops and the timing and best treatment(s) for stiffness are limited. A recent James Lind Alliance Priority Setting Partnership identified stiffness after TKR as a top-10 research priority to better understand and test interventions. Current measures are not accurate or suitable for use in the home. The investigators need tools to accurately measure early indicators for stiffness.
2. Rationale

The investigators currently have no tool to remotely and accurately detect development of early post-surgical knee stiffness. This study aims to develop a cost-effective tool to measure and quantify knee stiffness before and after total knee replacement (TKR) surgery for use across the NHS. The research seeks to understand how knee range of motion (ROM) recovers after TKR and detect early signs of stiffness. It also aims to predict who might develop stiffness after TKR and explore the relationship between pain and stiffness.

Current methods for measuring knee range of motion (ROM), such as hand-held tools for measuring angles, have limitations in terms of accuracy and need trained healthcare staff to use them. The ideal tool would be low-cost, easy to use, and provide rapid feedback to patients and clinical teams. The study will involve the development and validation of a computer vision-based approach (using cameras to assess movements) to monitor knee flexion and extension, and a walking pattern assessment. Video-based technology or computer vision (CV) has recently been pioneered in Exeter to measure spine movement in patients with ankylosing spondylitis. Computer vision is an emerging technology that has great potential for monitoring knee flexion in people with knee stiffness. This approach involves the use of cameras and machine learning algorithms to detect and analyse knee joint angles during movement automatically. By providing objective and accurate measurements of knee flexion, computer vision has the potential to improve the assessment of knee stiffness and facilitate targeted treatment interventions. However, as with any new technology, there is a need to validate the method in the context of patients with knee stiffness to ensure its accuracy and reliability. Studies have highlighted the importance of developing machine learning algorithms specifically for this patient population to account for individual differences in movement patterns and limitations due to stiffness. Further research is needed to assess the validity and feasibility of computer vision-based approaches for monitoring knee flexion in people with knee stiffness, which could ultimately improve the diagnosis, monitoring, and management of this condition.

Validation of the computer vision-based approach for monitoring knee flexion in people with knee stiffness is essential to ensure its reliability and accuracy. This requires developing and refining machine learning algorithms that can accurately detect and measure knee joint angles in this patient population. This study will evaluate the accuracy and precision of the algorithm against gold-standard measurement methods, such as motion capture or goniometry. Furthermore, this study will examine the sensitivity of the approach to changes in knee flexion due to stiffness and pain and assess its feasibility in a clinical setting. Once validated, the computer vision-based approach has the potential to provide a non-invasive and objective means of monitoring knee flexion in people with knee stiffness, which could inform treatment decisions and improve patient outcomes.

Another tool which the investigators will use is the Gaitcapture app which takes advantage of the accelerometer and gyroscope sensor in a mobile phone and acts similarly to an inertial measurement unit (IMU) to provide us with acceleration and rotation data.

The validation of the computer vision-based approach will involve comparing it against gold-standard measurement methods (specialist physiotherapy assessment).

In addition to the computer vision-based approach, the study will utilise body-worn sensors and mobile apps to monitor the physical activity levels, walking patterns and step counts of participants. This data will provide insights into people with TKR's overall physical activity patterns and help evaluate the usability, acceptability, feasibility, and accuracy of the tools for diagnostics and monitoring.

The findings of this research project have the potential to improve the diagnosis, monitoring, and management of knee stiffness after total knee replacement (TKR), with the potential to reducing the need for MUA surgery. By providing accurate measurements and early detection, the tools developed in this study could enable earlier referral to non-surgical treatments and reduce the need for costly and risky procedures to improve knee range of motion (ROM) after total knee replacement (TKR) surgery, like a manipulation of the knee under anaesthesia.

Here, the investigators will conduct a validation study of a marker-less motion capture algorithm to determine its accuracy and assess its feasibility and usability for implementation on a large scale in the home. the investigators will also ascertain the test-retest reliability of algorithm outputs such as knee flexion/extension angles.