Assessing the validity of inertial measurement units for shoulder kinematics using a commercial sensor-software system: A validation study

Abstract

Background and Aims: Wearable inertial sensors may offer additional kinematic parameters of the shoulder compared to traditional instruments such as goniometers when elaborate and time-consuming data processing procedures are undertaken. However, in clinical practice simple-real time motion analysis is required to improve clinical reasoning. Therefore, the aim was to assess the criterion validity between a portable “off-the-shelf” sensor-software system (IMU) and optical motion (Mocap) for measuring kinematic parameters during active shoulder movements. Methods: 24 healthy participants (9 female, 15 male, age 29 ± 4 years, height 177 ± 11 cm, weight 73 ± 14 kg) were included. Range of motion (ROM), total range of motion (TROM), peak and mean angular velocity of both systems were assessed during simple (abduction/adduction, horizontal flexion/horizontal extension, vertical flexion/extension, and external/internal rotation) and complex shoulder movements. Criterion validity was determined using intraclass-correlation coefficients (ICC), root mean square error (RMSE) and Bland and Altmann analysis (bias; upper and lower limits of agreement). Results: ROM and TROM analysis revealed inconsistent validity during simple (ICC: 0.040−0.733, RMSE: 9.7°−20.3°, bias: 1.2°−50.7°) and insufficient agreement during complex shoulder movements (ICC: 0.104−0.453, RMSE: 10.1°−23.3°, bias: 1.0°−55.9°). Peak angular velocity (ICC: 0.202−0.865, RMSE: 14.6°/s−26.7°/s, bias: 10.2°/s−29.9°/s) and mean angular velocity (ICC: 0.019-0.786, RMSE:6.1°/s−34.2°/s, bias: 1.6°/s−27.8°/s) were inconsistent. Conclusions: The “off-the-shelf” sensor-software system showed overall insufficient agreement with the gold standard. Further development of commercial IMU-software-solutions may increase measurement accuracy and permit their integration into everyday clinical practice.