I just stumbled across this 2013 thread earlier this month as we were finalizing/proofing a manuscript which I believe resolves the original questions/concerns raised. In contrast to some of the prior statements/conclusions -- in particular the ones about not using Prince et al. 1994 method to estimate prosthetic ankle-foot power -- I believe the Prince et al. 1994 method checks out (i.e., the math works out appropriately to estimate power due to deformation of structures distal to the landmark of interest). I would say that this method is one of the few preferred/recommended methods for estimating ankle-foot power in gait analysis (and quite important for prosthesis studies in particular). For details please see analytical derivations and experimental evidence presented in:

https://www.sciencedirect.com/scienc...21929018302902

In particular, please see Appendices B & C for full analytical derivations as to (i) why Prince et al. 1994 method is analytically equivalent to Distal Shank power (from Takahashi et al. 2012 Unified Deformable Model) when foot mass is negligible, and (ii) why Distal Shank power and Prince et al. 1994 method each provide reasonable/useful estimates of 6 degree-of-freedom power of the shank relative to the ground. Computing power due to motion between two bodies that are assumed rigid (e.g., shank-ground, or pylon-ground) allows for complete estimation of (prosthetic) ankle-foot power without the need to apply finite element methods from continuum mechanics (assuming foot mass & inertia are relatively small). Basically these methods enable us to estimate power due to deflection/motion of a prosthesis without having to directly model or make assumptions about the structure or geometry of the foot itself -- which is pretty cool!

PS. In this paper we also conclude/recommend that the gait analysis field should (in nearly all cases) stop using conventional inverse dynamics ankle power estimates (which compute power due to motion between a rigid shank and rigid foot, in which the foot in modeled as a single rigid segment and tracked via markers spread over the entire foot). This conventional inverse dynamics ankle power method can be highly problematic for studying prosthetic and biological limbs. We provide several empirical examples of how this common ankle power estimate can lead to large errors and incorrect conclusions related to science, technology development and clinical evaluation.