Fig 1.
Screenshot demonstrating feedback given to the Novices of the pilot study to indicate potential errors in marker placement. A lower-limb surface model was constructed and anatomical marker position data was overlaid to provide context for users (segment markers locations were identical for all trials, and therefore omitted). The marker position data consisted of expected positions (green circles), which were scaled to fit the surface model, and dimensionless IQRR scores (blue and red circles), which were scaled and positioned relative to the expected position with a connecting line to indicate directionality. Red circles indicated that a marker had crossed the threshold of 0.8 for the associated IQRR score, while blue circles indicated the marker was within the threshold. Participants were instructed to only modify those placements indicated by red circles, and to use their judgment in deciding how much to move the marker.
Fig 2.
Relationships between marker errors and changes in discrete variables.
Two basic relationships were observed: the left panel shows a highly linear relationship between anteroposterior marker placement error and kinematic change in the transverse plane, the right panel shows a non-linear relationship between vertical marker error and frontal plane kinematics.
Fig 3.
Tabulated results of the error sensitivity analysis.
Values shown are for right side kinematics only, however, left side trends were nearly identical. All values for kinematic change ratios represent a point estimate of the 95th percentile ratio, using the mean of 10 iterations of the simulation. All values shown are in normalized units of degrees of error per 10 mm of marker error (for angles), or degrees/s of error per 10 mm of marker error (for angular velocities). Empty cells indicate that a variable changed less than 0.5 degrees, or 5 degrees/second, for every 10 mm of error.
Table 1.
Tabulated results of the evaluation of the IQRR as a classifier of placement deviation, using maximum MCC.
Table 2.
Tabulated results of the evaluation of the IQRR as a classifier of placement deviation, using an IQRR threshold of 0.9 and finding maximum MCC.
Table 3.
Tabulated results of the evaluation of the IQRR as a classifier of placement deviation, using an IQRR threshold of 0.8 and finding maximum MCC.
Fig 4.
Bars indicate the signed median difference between the group of n = 6 Novices and the Expert, both before and after receiving feedback. * indicates significant differences from the Signed Wilcoxon Rank test (family-wise p <0.05).