Criterion for force distribution among muscles

[David Reichert]

Hello,
we're in the process of validating a model for the lower limb for rehabilitation purposes. In this context, we're comparing muscle force estimations of the Gait2392 OpenSim model to our model.
Subjects perform a simple squat exercise.
Like one would expect, the Vastus lateralis and medialis are the main muscles activated during the squat exercise.
Nevertheless, the Biceps femoris seems to be activated for knee stabilization.
(see figure attached, muscle forces for the Gait2392 model)

The question is: Which criterion does OpenSim use for the distribution of the knee moment among the muscles acting in knee flexion/extension as well as stabilization?

Thanks in advance,
David

[jimmy d]

There is information on how muscle forces are generated, including the equation for the objective function, here; https://simtk-confluence.stanford.edu/d ... tion+Works

[David Reichert]

Thank you for answering and sorry for the late reply.
What remains unclear to me is the following question:

Somehow OpenSim seems to assume a certain activation level for the Biceps femoris in order to minimize the objective function of muscle activations.
I don't see where this comes from. Looking at the objective function and constraints, for me it seems logical to generate the knee moment only by activating the Vastus muscles. In other words, I still don't understand what criterion is used to activate the Biceps femoris as well.
I hope I could make myself clear. Thanks for answering in advance.

[jimmy d]

There are other things that generate force, like passive properties of muscle and tendon. Biceps Femoris is also a two joint muscle (hip and knee), so just assuming that its action is concerned with just the knee is not correct.

The criterion for the optimizer to find activations is the link I posted before. There is a target acceleration for the model, and the optimizer is trying pick a set of activations for all the actuators which generates those accelerations while minimizing activation squared. This control law is not how we actually work, and you can add constraints on the activation if you feel the activations are too high or too low, that is why we recommend collecting EMG on muscles of interest.

I would suggest going through this paper, it is a good resource.

[David Reichert]

Thank you, that helped a lot

[cui jiashuo]

Hello, how did you get this picture? I also want to use deepsim for deep analysis. Can you give me some direction?