An interesting project but it still doesn't address the fact that mapping must be done between participating systems and I cannot imagine that their mapping tool is any better than HL V2 tools that have many years of development and usage experience embedded.

Nor does it address the issue that every federated approach has and that is the availability of the networks and systems that are under disparate control and allowable access to the information stored there.

Is it smart to lead a physician to think they have all of a patient's information when there is a good chance they do not? This is exactly what federated systems are being promoted as providing. This is especially so in the given (from the article) example of the auto accident.

It is implied that some magic has been worked on all systems at every facility where this patient could have ever been treated. In addition to them all being able to share information (given appropriate credentialing) they are all connected and available on the P2P network at all times. In many parts of the US and certainly the world this is truely an unrealistic expectation.

There are more realistic ways to deal with these issues.