FIGURE 2.
Videos were collected within a clinical trial. These videos were scored by three raters. From these videos, stick figures have been extracted. From the x,y coordinates of the body points, meaningful movement and position features are calculated and used to train a random forest regressor towards the clinical scores.