> One therapist is taking measurements of say your arm motion and making inferences about the motion of other muscles.
This seems like a sequential Bayesian filtering problem. Probably high enough dimension that you should just use a particle filter. The big seminal background text in this area is Bishop: Pattern Recognition and Machine Learning.
If the "motion of other muscles" is inferring pose, you could also look into what computer graphics calls inverse kinematics (a typical IK model has a number of dimensions that could fit into a particle filter). There's some more in-depth stuff in motion planning that actually takes into account muscle capability. But I wouldn't know where to find info on that, short of watching the last several years of Siggraph Technical Papers Trailers, grabbing all the motion planning ones, then reading everything they cite.
I've heard of inverse kinematics but I think it's more focused on "modeling" than statistics/probability? That is, you would have to model each muscle?
I think he is doing something that is more "invariant" across human variation? (strength, body dimensions, age, etc.) I'm not sure which is why my question was vague, but this is helpful.
Yeah, IK is about pose and motion modeling. But you can put any state+motion model inside sequential Bayes, and get the probability that the model is in a particular configuration out.
Hard to know whether that's relevant without knowing what he's trying to predict though.
This seems like a sequential Bayesian filtering problem. Probably high enough dimension that you should just use a particle filter. The big seminal background text in this area is Bishop: Pattern Recognition and Machine Learning.
If the "motion of other muscles" is inferring pose, you could also look into what computer graphics calls inverse kinematics (a typical IK model has a number of dimensions that could fit into a particle filter). There's some more in-depth stuff in motion planning that actually takes into account muscle capability. But I wouldn't know where to find info on that, short of watching the last several years of Siggraph Technical Papers Trailers, grabbing all the motion planning ones, then reading everything they cite.