Criterion validity of wearable sensors in lower limb joint angle estimation: A systematic review

Introduction/Background Motion capture systems are extensively used to measure human joint kinematics. These systems, however, lack portability and can only be used in laboratory settings. Recently, portable and wireless systems, such as inertial measurement units (IMU), have been developed and are seen as an alternative to quantify human movement outside of the laboratory. Several studies have established their psychometric properties. However, the variety of systems used and joints evaluated make it difficult to conclude on the validity of the different systems available. The aim of this systematic review is to determine the criterion validity of IMUs for the evaluation of lower limb kinematics when compared to motion capture systems. Material and method Five different databases were screened (Pubmed, Cinhal, Embase, Ergonomic abstract and Compendex). Included articles had to report on criterion validity of IMUs (vs. motion capture systems) of at least one joint in healthy adults. Only articles published after 2005 were considered. Pairs of raters conducted data extraction and critical appraisal using structured tools, and consensus had to be reached. Results A total of ten articles was included. Quality ratings of 30% of the studies reviewed were characterized as high, and 70% as moderated (interrater reliability: ICC = 0.7). Knee and trunk movements showed moderate to good criterion validity, while hip and ankle movements were characterized as having poor to good validity. Conclusion IMU systems are valid to evaluate knee and trunk kinematics. For the hip and ankle, however, the validity varies according to movement studied. The measurement of hip flexion/extension, abduction/adduction and ankle plantar flexion/dorsiflexion with IMU can be considered as valid. For ankle inversion/eversion and hip internal/external rotation, the validity is poor. More studies are therefore needed to optimize the use of IMU for these movements (e.g. different calibration or biomechanical models).