Gait analysis is an important clinical tool. A variety of models are used for gait analysis, each yielding different results. Errors in model outputs can occur due to inaccurate marker placement and skin motion artefacts, which may be reduced using a cluster-based model. We aimed to compare a custom-made cluster model (ClusBB) with Vicon's plug-in gait. A total of 21 healthy subjects wore marker sets for the ClusBB and plug-in gait models simultaneously while walking on a 6-m walkway. Marker and force plate data were captured synchronously and joint angles/moments were calculated using both models. There was good correlation between the models (coefficient of multiple correlations > 0.65) and good intra-session correlation for both models (coefficient of multiple correlations > 0.80). Inter-subject variability was high, ranging from 15° to 40° in the sagittal plane and 11° to 52° in the coronal and transverse planes. Intra-subject variability was small for both ClusBB and plug-in gait models. Inter-subject variance tended to be high in both models for knee abduction/adduction, but particularly so for plug-in gait, suggesting that a cluster-based model may reduce the variability. The inter-subject variance in out-of-sagittal plane data is of particular importance clinically, given the reliance on these datasets in clinical decision-making.

No abstract