I worked on a team that spent 5 years on a realtime data protocol for one of these companies and didn't get very far. I left 6 months ago and they're still working on it.
We are working on technology which begins to get at this locked-in data via an API package which would be installed at individual institutions. The technology utilizes HL7 messaging to get at data from the major players (ie. Epic, GE, Cerner). The problem is the main software vendor, Epic, has built their technology up, by erecting proprietary barriers, an API would kill their entire model. Therefore the best way to get at this industry is through the providers, as they are becoming increasingly burden by these proprietary, legacy systems.
From a high level perspective, "health 2.0" is very interesting to me, and the research I've done in recent months has led me to believe that it's not a place for your run of the mill startup. You need some serious gunpowder to survive. However, it does seem ripe for a few disruptions.
As was stated in a previous comment, there are a large variety of applications which would benefit from access to actual medical data, and not require manuel entry. IMHO, this manual entry is why services like Google Health and Microsoft HealthVault will never take off.
