simtk.org

Hello,

I would like to compare OpenSim results (muscle activation) to phisiological results (measured activations).

I'm using Arm26.osim and I had some results that seemed a bit weird: some muscles activations where too high (~0.8) for a long time, while the movement was so simple (flexion).

Yet in order to have a meaningful comparaison, I need to understand the biomechanical model (the used approximations, flv relation, muscle detailed model, etc.)

I already read the published paper as well as some of the papers cited in the bibliography but I couldn't find satisfying answers (I need details).
F.E.Zajac's paper "muscle and tendon: properties, models, scaling, and application to biomechanics and motor control" helped me to understand some things, yet I still don't know if it was used when designing OpenSim.

Thank you.

Hi Maha,

In no way I'd compare the OpenSim Arm26 model with a real model or draw scientific conclusions from it. The Arm26 model has very limited number of degrees of freedom (a double pendulum with 2 pins and 6 muscles) provided only for educational purposes. On the other end of complexity is the upper extremity model at https://simtk.org/home/up-ext-model (please read the documentation and publications for this one). Depending on your needs you may want to simplify the model quite a bit before using it.

If you have specific questions about the Arm26 or any other model distributed with OpenSim, please feel free to ask.

Best regards,
-Ayman

Hi Ayman,

I used the model you specified above.

We faced a problem with the inverse dynamics tool that we resolved by modifying the mass and inertial values of the bodies.
Then, when we tried to use the static optimization tool, the optimization failed, with the following error message: "the model appears to be too weak", we doubled the maximum isometric force value for the muscles, but the optimization did continue to fail, with a new error message: "the model appears to be unsuitable for the optimization". we locked 10 joints (wrist joints)and we left the others free (we are studying arm flexion).
Yet this didn't solve the problem, at each time step, we were asked to lock the other joints.
And, looking at the results (muscle activations), they seem too far from being realistic (in comparison with the EMG data we recorded), because some muscles are activated while they shouldn't be (e.g. triceps instead of biceps).
Is there something we could change to get normal results?

We're also wondering if someone has successfully run all the tools on this model and how they did solve these problems if they faced them.

Thanks a lot.

Maha.