Abstract:
There is a growing demand to move standard gait and musculoskeletal analyses towards outside of the laboratory. In this study, we aim to evaluate the use of inertial sensing devices as a surrogate tool of joint loading. Using open source musculoskeletal models, we have simulated joint kinematics, kinetics and condylar loads in OpenSim to evaluate the potential use of inertial sensing devices. Peak tibial accelerations at heel strike were correlated against peak joint kinetics and condylar loads during weight acceptance. Furthermore, partial least squares regression (PLSR) statistical analyses were conducted to determine any machine learnt relationship between these data. Using the PLSR analyses, a linear relationship in the prediction error against the simulated joint kinematics and condylar loads was found, indicating an absence of a predictive variable such as further anthropometric data. A linear correction factor was applied to the predictions to illustrate the missing variable(s), concluding that the scattering of the data was due to white noise.