Hi all
I'm curious about the validation of OpenSim musculoskeletal models.
Is it necessary to acquire subjects' joint moment data with dynamometer?
or alternatively, after
1.the segmental parameters being scaled,
2.residuals reduced,
3.the reserve actuators' forces eliminated by tweaking max isometric force through CMC process;
4.IK results fit with experimental data.
After these procedures, could we say the model is validated (without dynamometer data)?
Thanks!
Shiuan-Lin
Model Validation
-
Edith Arnold
- Posts: 44
- Joined: Fri Apr 06, 2007 2:07 pm
RE: Model Validation
Hi Shiuan-Lin,
Taking a subject's joint moment data with a dynamometer and adjusting the model's Fmax values would be a good way to validate a model if you want to argue that it is subject specific, but I think most people do not take this step. Rather, they follow a procedure more like the one you've outlined here. Scaling the segments adjusts the moment arms of the muscles, then they make sure the muscles are strong enough to reduce the need for reserve actuators.
One common phenomenon though, is that a model that is strong enough to reduce the contributions from reserve actuators will be much much stronger than dynamometer results would indicate.
My preference in modeling is to make tweaking no more complicated than you have data to support. So, in my mind, the answer to your question is a qualified "yes." If I didn't have dynamometer data telling me something about a subject's joint moments relative to population data (e.g. this person has proportionally stronger knee extensors, but weaker ankle dorsiflexors) I would only adjust Fmax with a constant scaling factor. If I tweaked every muscle individually I would want some experimental data to back up my decisions, otherwise I would say it wasn't validated.
Those are my thoughts. If the relative strengths of the muscle groups vary meaningfully among subjects, a tool to easily take a few dynamometer measurements from a subject and adjust the model quickly would be really useful.
-Edith
-Edith
Taking a subject's joint moment data with a dynamometer and adjusting the model's Fmax values would be a good way to validate a model if you want to argue that it is subject specific, but I think most people do not take this step. Rather, they follow a procedure more like the one you've outlined here. Scaling the segments adjusts the moment arms of the muscles, then they make sure the muscles are strong enough to reduce the need for reserve actuators.
One common phenomenon though, is that a model that is strong enough to reduce the contributions from reserve actuators will be much much stronger than dynamometer results would indicate.
My preference in modeling is to make tweaking no more complicated than you have data to support. So, in my mind, the answer to your question is a qualified "yes." If I didn't have dynamometer data telling me something about a subject's joint moments relative to population data (e.g. this person has proportionally stronger knee extensors, but weaker ankle dorsiflexors) I would only adjust Fmax with a constant scaling factor. If I tweaked every muscle individually I would want some experimental data to back up my decisions, otherwise I would say it wasn't validated.
Those are my thoughts. If the relative strengths of the muscle groups vary meaningfully among subjects, a tool to easily take a few dynamometer measurements from a subject and adjust the model quickly would be really useful.
-Edith
-Edith
-
Shiuan-Lin Huang
- Posts: 7
- Joined: Mon Mar 03, 2008 7:18 pm
RE: Model Validation
Thank you, Edith.
These informations are really helpful, clarifying the doubt about tweaking Fmax.
Shiuan-Lin
These informations are really helpful, clarifying the doubt about tweaking Fmax.
Shiuan-Lin