Here is the presentation that Dr. Paul Veregge and I gave at TEPR 2006. Paul did all the work (thanks Paul):
Data Mining for Disease Management and Patient Communication
I call it the Poor Man’s EHR. It demonstrates how you can achieve many of the benefits of an EHR by data mining your existing systems. It is a great approach if you don’t have $60M in your pocket.
Candid CIO 
I was reading your BLOG and came across this previous entry. I two think that all of the data for an E.H.R. system is available in our current environments but it is difficult to access. At SJH, we replicate patient care system data to a SQL server in order to access it but data that passes to and from external systems is not contained in that repository. That is why I am a proponent of using the messages that daily flow through our integration engine as a source for the E.H.R. data. That data flows through in near real-time and is normalized to the HL7 or EDI standards so if we captured data from all of our systems it could be reconciled and placed into data analysis cubes. Some integration engines do this better then others but most engines with some additional programming effort could perform this task.
As you may know, I am a proponent BizTalk 2006 with the HL7 and HIPPA accelerators as an integration engine because it not only is the best technology solution for integration but also supports the infrastructure to allow messaging data to be extracted to be used in analysis and monitoring of real-time medical processes.