Assessment of wearable IMU devices for monitoring recovery following TKA surgery

Tools

Ligeti, Alexandra and Forsyth, Lauren Emma and Clarke, Jon and Riches, Phil (2025) Assessment of wearable IMU devices for monitoring recovery following TKA surgery. In: Conference on New Technologies for Computer/Robot Assisted Surgery, 2025-09-10 - 2025-09-12, Portugal.

[thumbnail of CRAS-2025-Assessment-of-wearable-IMU-devices-for-monitoring-recovery-following-TKA-surgery]
Preview
 Text. Filename: CRAS-2025-Assessment-of-wearable-IMU-devices-for-monitoring-recovery-following-TKA-surgery.pdf
Accepted Author Manuscript
License: Strathprints license 1.0
Download (269kB)| Preview

Abstract

INTRODUCTION Computer Assisted Orthopaedic Surgery (CAOS) has been available for over two decades, yet its widespread adoption remains limited. Navigation and robotic systems have steadily evolved, offering enhanced accuracy, precision, and reproducibility [1]. However, definitive evidence linking these technologies to long-term functional benefits is still lacking [1]. The orthopaedic community's cautious uptake is further influenced by concerns around cost-effectiveness, limited third-party reimbursement, and industry-driven promotion rather than patient-centric data. Wearable technologies present a solution to bridge this gap, particularly in joint replacement surgery. By enabling continuous and remote monitoring of patient function preoperatively, intraoperatively, and postoperatively, wearables offer the potential to optimise surgical planning and postoperative recovery, subsequently leading to better functional outcomes. In total knee arthroplasty (TKA), range of motion (ROM) is a critical determinant of walking ability, daily function, and is associated with patient satisfaction [2, 3], early post-surgical gains are especially important. Data-driven insights from wearable sensors can assist in tailoring surgical decisions, improving rehabilitation strategies, and ultimately enhancing functional outcomes. For these technologies to be successfully integrated into clinical and home-based rehabilitation pathways, they must provide accurate, reliable, and actionable information. This is particularly vital as healthcare trends shift toward remote care, personalised interventions, and cost-effective solutions that minimise revision surgeries and complications. The aim of this study was to assess knee range of motion using the MotionSenseTM sensors and validate the results against a clinical motion capture standard in a TKA patient pre- and post-TKA surgery. MATERIALS AND METHODS Ten TKA patients aged between 53 - 71 years old were recruited for this study (Weight: 88.0 ± 15.6kg); Height: 1.73 ± 0.12m; BMI: 30.09 ± 3.22kg/m2, mean ± standard deviation). Participants are expected to visit the gait laboratory a total of 3 times, once preoperatively and twice after surgery; 1 week, and 6 weeks. Movement was analysed using Vicon Bonita cameras to track retro-reflective markers that were attached to the lower body as per the PlugInGaitTM model. In addition, MotionSenseTM sensors were attached unilaterally to the lower limb (Figure 1). The participants completed 3 activities; treadmill walking, stair ascent/descent and maximum standing flexion/extension. Treadmill walking consisted of 5 minutes walking at a self-selected comfortable speed on a level treadmill. Using a bespoke graphical user interface, following a ~1 minute habituation period the first 10 gait cycles per participant were manually isolated, with the population mean gait cycle analysed. The stair ascent/descent was completed 3 times per trial and an average ascent/descent determined for individual patients and a population average evaluated. The standing flexion/extension activity included 3 repetitions per participant of a standing maximum flexion and extension movement, the pooled population range of motion was assessed. Knee flexion was determined by Vicon PlugInGaitTM (100 Hz) and by a proprietary algorithm in a mobile device to which the MotionSenseTM sensors (~50Hz) exported data in real-time. Following up-sampling to 100Hz, cross-correlation was used to time synchronise the measurements in gait cycle windows identified from peak flexion to peak flexion. As the zero point for knee flexion depends on marker placement, the mean knee flexion was subtracted from each data set before calculating a root mean square error (RMSE) between the technologies, determined in each gait cycle window. This analysis was repeated for the stair climb and descent activity, as well as the maximum standing flexion/extension movement. RESULTS Preoperatively, the TKA cohort walked at a speed of 0.56 ± 0.14 m/s. One week postoperatively, this decreased to 0.52 ± 0.14 m/s, and by 6 weeks postoperatively, it increased to 0.60 ± 0.28 m/s. No statistically significant differences in walking speed were observed between sessions (p > 0.05). Knee ROM was similar during both walking and stair ascent/descent but decreased following surgery. No data were available for the standing flexion/extension activity at one week postoperatively due to pain and swelling. During walking and stair navigation tasks, participants remained in a flexed position, showing limited ability to fully extend the knee post-TKA. However, by 6 weeks postoperative, ROM had improved across all activities compared to the 1 week postoperative time point (Figure 2). Before surgery, RMSE and signed difference data demonstrated excellent agreement between the MotionSense™ sensors and the Vicon system across all activities (Tables 1 and 2). Interestingly, RMSE values were slightly lower at 1 week postoperatively compared to preoperative and 6 week postoperative measures (p > 0.05), possibly due to reduced movement speed and ROM resulting from pain and swelling. Nevertheless, RMSE results were consistent with previously reported research, demonstrating strong correlations between the two technologies. CONCLUSION AND DISCUSSION MotionSense™ sensors performed accurately across all activities. As expected, one week post-TKA, knee ROM was limited, which may have contributed to the higher accuracy observed. However, all results were found within clinically acceptable thresholds These findings highlight the potential of wearable technologies to provide precise and comprehensive data to inform patient-specific surgical planning and personalised rehabilitation strategies. By enabling continuous, granular monitoring of functional recovery, such technologies have the capacity to support more effective decision-making and ultimately improve functional outcomes in orthopaedic care.

ORCID iDs

Ligeti, Alexandra, Forsyth, Lauren Emma ORCID logoORCID: https://orcid.org/0000-0002-9520-8984, Clarke, Jon and Riches, Phil ORCID logoORCID: https://orcid.org/0000-0002-7708-4607;
CORE (COnnecting REpositories)