Biomedical Engineering Reference
In-Depth Information
Boxlifting R
2
Hip
R
2
= 0.9842
100
80
Knee
R
2
= 0.978
60
40
Ankle
R
2
= 0.8945
20
Torso
R
2
= 0.8212
0
Experiment
-100 -80
-60
-40
-20
0
20
40
60
80
100
-20
Shoulder Phase 1
R
2
= 0.9938
-40
Shoulder Phase 2
R
2
= 0.3982
-60
-80
Elbow
R
2
= 0.7696
-100
Simulation
FIGURE 9.20
The R
2
values for experiment versus predictive simulation.
flexion of 12 subjects. Other symbols in
Figure 9.20
represent the relation for the
rest of the determinants.
9.6.4.3
Discussion
As indicated in
Figure 9.5
, the validation framework is based on four benchmark
tests comprising two qualitative and two quantitative benchmark tests. The results
for the first qualitative benchmark test (which demonstrates subjective compari-
sons by observing videos of the simulated and experimental models shown in
Figure 9.15
) have shown a reasonable correlation between the experimental and
the predictive models with an r-value of 0.692729.
The subjective evaluation may not give information as accurate or specific as
that obtained by the objective evaluation, but it can give some expectations to the
modeler that the objective measure cannot provide. For example, the subjective
measure can present how realistically the whole body is moving during the task.
It also presents images that may help in finding or localizing unnatural behaviors
at certain joints that could be hard to observe by just looking at the data coming
from the objective evaluation. The authors found that the subjective assessment
was very useful for the modeler, especially when the motion is far from natural,
or the strategy of the task is different from what the model predicts. While the
subjective assessment gives general information, it can still point out specific pro-
blems that can inform the quantitative assessment. For example, major differences
in the whole-body motion will make the modeler think about a different objective
Search WWH ::
Custom Search