Model Validation Question

[Nicholas Vandenberg]

Hey all,

I had a more random question as I'm nearing the end of a project and thinking about validation of baseline solutions tracking experimental data. I know from (Hicks, et al., 2015) that a method for validation is a minimization of residual/reserve forces and moments, or at least getting them below a specified threshold. Based off what I've seen on the forum, I typically have added weak reserve actuators for each DOF in a muscle-driven simulation. I'm wondering if the output from these actuators could be considered as analogous to these same reserves from a typical OpenSim workflow or not?

Best,
Nick

[Ross Miller]

I haven't done the old IK-->RRA-->CMC for a while, but I think the reserve actuators in that workflow are conceptually similar to what would be done in a MocoTrack or "predictive simulation" problem (non-muscle actuators given a large weight to discourage the optimizer from using them unless really necessary).

I prefer to not use reserve actuators at all in Moco because I'm never sure how small is "small enough" (easier to validate things like muscle mass and joint strength), but there's of course nothing conceptually wrong with using them.

Ross

[Luke Jessup]

We've been playing around a lot with different reserve actuation strengths with MocoInverse on graded walking conditions. Our validation involved minimising those residual forces below some threshold, and we luckily had some experimental emg, fascicle length and metabolic data to compare to. We'll probably be using a max strength between 5-10 Nm / N for these reserves, mainly to provide a little assistance around the hip. We didn't want to go above that value because they do have a pretty big influence on the solution, and like Ross said, it's hard to know how small is small enough. Based on other posts on here and the opensim forum though, it seems the value that you decide on is very much case-by-case.

Luke