ABSTRACT
The assessment of coordination variability in multi-joint human movements has traditionally started from angle-angle representations, and then used the angle change between subsequent time points as input for further analysis through vector coding. We propose an improvement to this approach, and suggest employing angular velocities as input data (Velocity Ellipse Method, VEM). We used experimental data and theoretical principles to contrast VEM with an existing standard (Difference Ellipse Method) and discuss its advantages and potential issues. Normalised cross-correlation was used to compare VEM and DEM in 36 angle couplings, from 20 participants running at 12 km/h on a treadmill. The hip flexion/extension-knee flexion/extension data were further investigated to discuss the robustness of the approach to measurement noise and outliers. Although DEM and VEM generally exhibited similar patterns (cross-correlation between 0.851 and 0.999), the variability curves from the two methods were noticeably different in some intervals. Also, using angular velocities as input appeared more robust to potential noise from raw data whilst retaining the following features: (a) more coherent with biomechanical conventions for calculating three-dimensional angular dynamics; (b) still suitable for coordination analysis; and, (c) more easily interpretable by practitioners when represented as relative motion plots.
