CASP is all for structure prediction. This has really nothing to do with things like Alzheimer's, Mad Cow, etc. These diseases are caused by protein misfolding.
Thus, unlike CASP, we are trying to understand something bigger -- not just what is the final shape, but *how* proteins reach the final shape. This is really the critical piece of info for tackling protein folding related disease.
With that said, we actually have discovered a new way to use F@H-like data for structure prediction. While that's not our primary goal, this work has been precirculated to our colleagues running CASP and they have used the info in their methods. Our paper describing this new method should be coming out in Journal of Molecular Biology pretty soon (was recently accepted).
Vijay