How can I make torque-driven predictive simulations realistic?

[Nicholas Bianco]

Thanks for linking that Tylan, and glad it works well for you. It seems like we should add this actuator to the main OpenSim codebase given the level of interest here.

My question is regarding the polynomial coefficients that are used in the Ashby Model. Should these be the same regardless of movement? And is there any guidance on how to generate these coefficients for additional degrees of freedom?

I don't know enough about the original Ashby model to fully answer this question unfortunately. The best resource for that is to go directly to Blake Ashby's thesis from 2004. But my best guess/recommendation would be to treat these curves like generic muscle force-length/force-velocity curves. They may not be necessarily accurate to specific muscle groups, but they are a better approximation to the torque-generating capacity of a muscle-driven model compared to the default CoordinateActuator.

Best,
Nick

[Ross Miller]

This paper has some great data on torque-angle and torque-angular velocity data in the hip, knee, and ankle (sagittal plane only):

https://pubmed.ncbi.nlm.nih.gov/17485097/

Table 2 gives all the data needed to reconstruct the plots in Figure 3. I use this paper a lot to check if my model has normal strength or Hulk strength but I think it would also be useful for parameterizing joint-level actuators with force-length and force-velocity effects.

Ross

[Aaron Fox]

Thanks for the link to the paper Ross. I suspect there might be some data out there somewhere for hip strength across the frontal/transverse plane, but I doubt there would be at the knee? I'm keen to include frontal and transverse knee torques but am a little unsure of how to program these.

FYI I'd be keen to write a paper with a Hulk strength model

Aaron

[Ross Miller]

Markolf et al. (1976) is the "classic" reference on passive torque-angle and force-displacement relationships at the knee in 3-D. I've not seen a paper that considers "active" joint-level relationships in those DOF.

https://pubmed.ncbi.nlm.nih.gov/946969/

[Aaron Fox]

Thanks for that Ross - it makes sense why there is no data as I don't know how I'd even try to produce an active valgus/varus torque at the knee myself?

I suspect this would produce unrealistically high values, but do you think running simulations with the model and moving the joint through various degrees of motion with the muscles fully activated (i.e. simulating isometric or isokinetic type dynamometry) would give something reasonable?

Aaron

[Pasha van Bijlert]

Hi Aaron,

Not completely sure why you want controllable valgus/varus torques, but something you could try: take any converged optimization you may have where the knee joint is a perfect hinge, and plot the joint reaction torques (M_x, I guess). This may give you (order of magnitude wise) an indication of the torques you want to implement. I used this to parametrize a joint spring on a DOF that was previously not free (I went from a hinge to a biaxial joint), although I ended up needing a lower max torque than the peak reaction moments.

Cheers,
Pasha

[Ross Miller]

Pras Sritharan has a nice paper on this issue:

https://onlinelibrary.wiley.com/doi/abs ... /jor.22082

You can see from Fig. 4 that muscles spanning the knee (which is what we'd be modeling by including an active torque-generator in this plane) do make substantial contributions to the net frontal plane moment at the knee (Fig. 4a) although it's relatively small compared to the "muscles that don't span the knee" contribution.

Ross

PS2012.jpg (328.78 KiB) Viewed 766 times