Wearable sensors allow remote monitoring of patient recovery after knee replacement

A two-sensor wearable inertial measurement unit (IMU) system can reliably measure knee flexion and facilitate remote monitoring of recovery among patients who had undergone total knee arthroplasty (TKA), according to a recent study.

“In this study, various aspects contributing to the accuracy and repeatability of an IMU-based wearable system in measuring knee flexion have been evaluated. Based on the data collected from a total of eight patients that recently underwent TKA, the accuracy of the wearable system for measuring knee flexion was found to be below the desired five-degree threshold,” the researchers said.

Eight patients who had just undergone TKA were enrolled. All were given the MotionSense wearable IMU device, which consists of two sensors attached laterally above and below the knee joint, as well as a Bluetooth-connected app that patients installed on their phones. Two knee flexion references were used: static goniometer readings by a physiotherapist and dynamic machine learning estimations from video motion capture data.

Goniometer readings were collected both at baseline and at the 10-day follow-up, during which the physiotherapist was blinded to the readings from the previous visit. Measurements for the whole cohort ranged from 50 to 74 degrees. The mean absolute within-patient difference in flexion angle was 2.6±2.7 degrees. [J Funct Morphol Kinesiol 2021;6:60]

For the device measurements, patients were instructed to place their operated leg on top of a box, a process repeated five times and performed on both sessions. The resulting mean variability in measurements was 1.6±1.5 degrees.

To assess device reliability while moving, patients were asked to perform long arc quad and standing knee bend exercises. Due to faulty measurements by the machine learning reference, data from three participants were excluded. The remaining five logged a total of 22 exercise sessions.

After removing overall bias in measurements, the IMU device had a mean absolute error through the range of motion of 2.6 degrees, as compared with the motion capture reference. Error was greater in deep flexion and towards terminal extension, but a piecewise linear correction method could be applied to further minimize such deviations. The final mean absolute error through the range of motion was 2.3 degrees.

In total, the combined mean absolute error was 4.2 degrees without applying corrections. According to the researchers, “this compares well to current clinical practice, where goniometer measurements or visual assessments remain the standard of care.” Such visual evaluations could deviate by up to 5 percent.

“Based on the promising results from this pilot study, using the presented system to accurately monitor patients’ knee flexion angle in a remote setting will be explored in future work,” they added.