From: A guide to inter-joint coordination characterization for discrete movements: a comparative study
Metrics | Recommendations |
---|---|
Temporal coordination | – |
Zero-crossing time interval | – |
Inter-joint coupling interval | – |
Angle ratio | – |
Angle–angle plot | \(\bullet\) Same limits on axis to avoid noise zooming |
\(\bullet\) Using position data enhances better spatial coordination strategy | |
\(\bullet\) Using velocity data erases differences due to different starting positions | |
\(\bullet\) The ratio of the widths of the point distributions highlights the coordination pattern strategy | |
\(\bullet\) The area covered by the point shows the temporal coordination strategy | |
\(\bullet\) Coupling this metric with other indicators (as the Angular Coefficient of Correspondence for example) helps to reduce its dimensionality | |
Continuous relative phase | \(\bullet\) Normalize data at the range |
\(\bullet\) Unwrap the result to get a meaningful MARP | |
Correlation coefficient statistics | \(\bullet\) Compute with data position |
\(\bullet\) Always check the p-value, if too high, the result is not interpretable | |
\(\bullet\) Spearman or Pearson Correlation Coefficient | |
\(\bullet\) Use a low-pass filter on data first | |
\(\bullet\) Use a threshold on data’s velocity to remove micro-movement (if the joint velocity is below the threshold, joint trajectory is considered constant) | |
Cross-correlation | \(\bullet\) To compute on velocity data |
Really sensitive to targets’ position | |
Relative joint angle correlation | – |
Principal component analysis | \(\bullet\) To compute on velocity data |
Distance between PC | \(\bullet\) To compute on velocity data |
Atypical kinematics | \(\bullet\) To compute on velocity data |
\(\bullet\) Needs a very large amount of data |