Prescribing joint co-ordinates during model preparation in OpenSim improves lower limb unplanned sidestepping kinematics.

OBJECTIVES: Investigate how prescribing participant-specific joint co-ordinates during model preparation influences the measurement agreement of inverse kinematic (IK) derived unplanned sidestepping (UnSS) lower limb kinematics in OpenSim in comparison to an established direct kinematic (DK) model. DESIGN: Parallel forms repeatability. METHODS: The lower limb UnSS kinematics of 20 elite female athletes were calculated using: 1) an established DK model (criterion) and, 2) two IK models; one with (IKPC) and one without (IK0) participant-specific joint co-ordinates prescribed during the marker registration phase of model preparation in OpenSim. Time-varying kinematic analyses were performed using one dimensional (1D) statistical parametric mapping (α=0.05), where zero dimensional (0D) Root Mean Squared Error (RMSE) estimates were calculated and used as a surrogate effect size estimates. RESULTS: Statistical differences were observed between the IKPC and DK derived kinematics as well as the IK0 and DK derived kinematics. For the IKPC and DK models, mean kinematic differences over stance for the three dimensional (3D) hip joint, 3D knee joint and ankle flexion/extension (F/E) degrees of freedom (DoF) were 46±40% (RMSE=5±5°), 56±31% (RMSE=7±4°) and 3% (RMSE=2°) respectively. For the IK0 and DK models, mean kinematics differences over stance for the 3D hip joint, 3D knee joint and ankle F/E DoF were 70±53% (RMSE=14±11°), 46±48% (RMSE=8±7°) and 100% (RMSE=11°) respectively. CONCLUSIONS: Prescribing participant-specific joint co-ordinates during model preparation improves the agreement of IK derived lower limb UnSS kinematics in OpenSim with an established DK model, as well as previously published in-vivo knee kinematic estimates.