"Space is hard" and this is harder, because it's about ourselves.
This has no chance until someone like Pres. Kennedy chooses to do it "because it is hard" to do.
I think this, updating legacy system (something I do), is close to the second problem, which is that no one person or team has a grasp on how it works now, and what ever is put in place will be in the exact the same situation: no person knows it all. The only way I've seen this addressed is that after the Big Problem is cut into the Smaller Problems, the people involved with each Small Problem have to be to only ones who understand it: it must be accepted that there will never ever be a global "this is how it works" person or team.
Then the first problem can be addressed: interoperability. Each Small Problem can then announce (or publish) how they solve their problems in their area of expertise. Then tell all the other vendors that they will interoperate with them in that way. This is not the way this has been stated so far, where the new system will interoperate to all the others, which won't happen for the simple fact that all the other vendors don't want to, its not in their self interest.
The third problem is inherent in our species: we change, so no static model can be used to organize the data about us. The next versions of HIV, Ebola, anti-Vaxers, are all changes that will strain any extensibility, but most especially a government/military/project planning approach that can only deal with statically defined, long term goals. In the 18 years of this project they will always play catchup.
So, yes, centrally defining EMR is going to fail, absolutely (or the Soviets would still be a power...).
My only poor suggestion is to find a way to give each medical specialty, including ones yet thought of, a way to define their own representations of their expertise and publish it so those working in that field can interoperate with each other. Something like a distributed/grid processing system that includes distributed people being involved.
Google searching still requires that you know that you have the results you want when you see it, where a library is organized to begin with, so Library Science has been trying to do this for a centuries and it might be worth studying to understand some of the problems.