Hi All,
I want to use OpenCap to collect data on dynamic sitting balance in people with quadriplegia. Very few of them will be able to punch their hand in the air. What are the possible alternatives for synchronization?
Possibilities include (are any of these viable)
- Assisted movement by a therapist (but somehow not being taller, e.g. kneeling)
- Different movement - head movement would be ideal (or punch elbow sideways)
- different method altogether??
We haven't set up the system. First wondering if it will work at all.
The task will be leaning the trunk in different directions and we want to be able to detect differences (accuracy) to 3-5cm - is that feasible?
thanks in advance
Maya
OpenCap - Alternative synchronization task??
-
Maya Panisset
- Posts: 1
- Joined: Fri Jun 21, 2024 6:35 pm
-
Matt Petrucci
- Posts: 229
- Joined: Fri Feb 24, 2012 11:49 am
Re: OpenCap - Alternative synchronization task??
Hi Maya,
You might want to just try collecting some pilot data with someone doing the leans to see if you run into any synch issues. The biggest issues usually occur for cyclic movements (like walking or biking) where the code may not know what cycle to sync together.
If you do need a hand punch, you might need to do something that is up as it is looking at the wrist and shoulder positions, and there needs to be some velocity signal in it as well (https://github.com/stanfordnmbl/opencap ... r.py#L1252). So your first option may be the best.
I think the closest tasks that was validated may be sit-to-stand, which you can see the mean absolute errors in Table 1 (https://journals.plos.org/ploscompbiol/ ... bi.1011462).
Hope this helps
You might want to just try collecting some pilot data with someone doing the leans to see if you run into any synch issues. The biggest issues usually occur for cyclic movements (like walking or biking) where the code may not know what cycle to sync together.
If you do need a hand punch, you might need to do something that is up as it is looking at the wrist and shoulder positions, and there needs to be some velocity signal in it as well (https://github.com/stanfordnmbl/opencap ... r.py#L1252). So your first option may be the best.
I think the closest tasks that was validated may be sit-to-stand, which you can see the mean absolute errors in Table 1 (https://journals.plos.org/ploscompbiol/ ... bi.1011462).
Hope this helps