Convert your FHIR JSON -> XML and back here. The CDA Book is sometimes listed for Kindle here and it is also SHIPPING from Amazon! See here for Errata.

Tuesday, July 10, 2012

Are Documents Dead?

I asked Gary Thomson over at CLOUD to take a look at the IHE mHealth profile.  He response asked "What if there aren't Documents?"  He then goes on to say
A piece of information does not know if it is medical, financial or educational. It also doesn't know if it is in an EHR or a PHR. It is just data, and if we wait to combine that information/data into a document until the last possible moment, then suddenly, the whole problem takes on a different view.
So even though he says, what if there are no documents, guess what?  There are.  They are just crafted dynamically.  IHE IT Infrastructure developed a "On-Demand Documents" variant of XDS/XCA that does just the sort of thing that Gary asks for, waiting to combine the data until it is requested.  The challenge later is that systems may need to reproduce exactly that content in the future.  So, we save it as a document.

The IHE mHealth Profile can interact and integrate with this "futuristic" concept of "document" just as readily as it can with today's concept.

About 4 years ago, the HL7 Structured Documents Workgroup developed a Structured Documents specification.  Now overtaken by events (CDA R3), this specification asserted that there five characteristics of documents that were generalized from the 6 characteristics of CDA documents:

  1. Human Readable
  2. Persistent
  3. Managed
  4. Contextualized
  5. Wholeness
Looking at this from the CLOUD perspective of "it's just data", I can see where three of these attributes apply.  A datum is readable, persistent, and managed.  By itself, it has no context.  In fact, as Gary says, it doesn't know if it's medical, financial or educational, or where it is.  That is all about context.  Google data without context and see what you get: Dirt.  Noise.  Meaningless.

To give data context, you must put it together and organize it.  And when you organize it as whole, you give it meaning.  And now that you have a whole, organized collection of data, what do you want to call it?  I'd like to continue calling it a document.

There's nothing about documents that says you cannot extract data from them to create new documents.  We do that every day.  This blog post did that from at least eight different documents.

My take-away from Gary's comments (which are good ones), is that the organizing principles used to obtain the "documents", or set the context, that the IHE mHealth Profile is asking about are where we need to focus.  Setting the context provides "data about data", otherwise known as metadata.  John Moehrke has a great post about healthcare metadata.

So, my one question about the IHE mHealth profile that is prompted by Gary's post is this:  Did we get the metadata right in the queries?  As I look through the query parameters, I see that we have John's six different metadata categories covered.  Where the profile is weakest is on "Exchange metadata", and I think that's a forgivable weakness.  Patients and providers aren't interested in organizing content around exchange metadata except for some rare cases (e.g., operational metrics).

I think we are in a broad sense, in good shape with this profile.

And to answer my question"  Nah, documents aren't dead.  In fact, in the new world order, they just might be a bit more "living".


  1. Shades of the PCAST report...

  2. People communicate through stories. Once upon a time, when humanity had no ability to record anything, there were no persistent stories (documents). The data elements for each story were stored in the bio-computer we call brain, all nicely tagged and organized, and were assembled dynamically when the owner of the brain would endeavor to tell the story. Changes to chemicals here and there caused the story to never be rendered exactly the same as before - both the data and the queries varied unpredictably. Carving stones changed the paradigm by persisting the results of one particular query into one particular brain (or a series thereof).
    If computers were bug free, theoretically, we could persist queries and data forever and recreate the stories at will, exactly as they were rendered at any particular moment in the past. I don't think we have enough storage in the galaxy for something like that, not to mention the technology tools. It is a compelling vision though to have all human brains supplemented by computers and interconnected with infinite memory and infinite accuracy at all times. Something like the Borg hive, only much better.