Page 1 of 1

Simulated Muscle Weakness

Posted: Tue May 03, 2022 10:51 am
by maddiegrosklos
Hi everyone,

I want to simulate hip flexor weakness during gait using Moco and I have marker data and EMG signals to use as input. I also have MVIC force production data from hip muscles.
I am basing the feasibility of my design off of the OpenSim Moco publication, specifically the section "Tracking a walking motion with normal and weakened muscles"

It states "We next weakened the ankle dorsiflexor muscles [tibialis anterior, extensor digitorum longus, and extensor hallucis longus] by reducing max isometric forces by 95% and solved the tracking problem again [using the "normal" solution for the initial guess]." "...We used the original cost terms that penalized squared controls and tracking error."

Has anyone applied such practices and have Matlab code they would share so I can visualize more details about how this is accomplished? I plan to track both marker data and EMG in the "normal" solution, but not use the EMG in the weakness conditions as this is what I want to alter. Is it appropriate to input the MVIC force values at all or does OpenSim compute these?

Has anyone applied this in a MocoStudy predictive simulation? Wouldn't a predictive solution technically be the most robust way to analyze?

Thanks in advance, I am trying my best with the documentation available but figured experience from someone who has done it might help me get over my hurdles.

Re: Simulated Muscle Weakness

Posted: Tue May 03, 2022 4:15 pm
by aafox
Hi Madeline,

I think you're on the right track with your thinking and approach. There's a few existing studies that have simulated the idea of muscle weakness - I worked on one from a cerebral palsy perspective (https://pubmed.ncbi.nlm.nih.gov/29353741/) and Marjolein van der Krogt performed a fairly broad general study of muscle weakness in gait (https://pubmed.ncbi.nlm.nih.gov/22386624/). I think you'll find that all of these studies simulate muscle weakness in the same way - in that they reduce the maximal isometric force of the muscles.

A limitation of these studies is that they have typically kept kinematics consistent, so the question was more so what control strategies maintain normal gait with weakened muscles. I agree with what you're suggesting that a predictive approach is a more robust way to look at this - as the question then becomes how does human gait potentially adapt with muscle weakness. There are some predictive gait examples that come built-in with Moco (e.g. the 2D walking) - my recommendation would be to review these and then adapt them to your scenario. For example, I think you could quite easily replicate the 2D walking example and simply use a model with reduced isometric forces and see what happens.

Aaron

Re: Simulated Muscle Weakness

Posted: Wed May 04, 2022 10:02 am
by nbianco
Hi Madeline,

You can find all the code from the paper examples here: https://github.com/stanfordnmbl/mocopaper. It's all in Python, but the API syntax is nearly the same between Matlab and Python, so it shouldn't be too hard to recreate the paper examples in Matlab.

Best,
Nick