SMART, VistA, and the Semantic Web

On the subject of VistA and the web of data, here's some background about SMART, a project to enable "substitutable" apps that run on multiple electronic health records systems, personal health records, data mining platforms, and health information exchanges.  We define a RESTful API to expose patient-level data using W3C's Resource Description Framework  -- and the WorldVistA team has built a prototype capable of running SMART Apps running on VistA!

SMART WorldVistA at NHS Hack Day 2012
http://www.smartplatforms.org/2012/06/uk-collaborators-build-smart-proof-of-concept-at-nhs-hack-day/

Developer documentation of the API, including our RDF data models
http://dev.smartplatforms.org

Background about SMART
http://smartplatforms.org

Comments

VistA's data model vs SMART's (and the VPR)

conor dowling's picture
one really nice thing about SMART is that here's a well thought out, general-purpose, designed-for-the-web model for patient care data, one that reuses standard vocabularies (VCARD etc) where it can and only when it has to does it introduce new forms of expression. It's hard to treat anything in isolation and VistA data is no exception. With this model, SMART presents a really nice opportunity for examining VistA's data. What parts of VistA's data model (vs == VistA Schema) have equivalent or related expressions in SMART? What's missing from either? How much of any transformation could be done through mapping declarations? What model translation is more involved? This also fits in nicely with another comparison we should do - how does VistA's schema compare to the VA's VPR "model". VPR is a roll up of VistA data the VA uses for CCD generation among other things. Unlike VistA's model or SMART's, VPR isn't formally defined but that shouldn't be hard to do. Two concrete comparisons - VistA Data Model to SMART, VistA Data Model to VPR - and we'd all have a much better view of VistA's native model of patient data.

data models and architecture

Tom Munnecke's picture

I might point out Mandel and Kohane's paper Escaping the EHR Trap — The Future of Health IT Kenneth D. Mandl, M.D., M.P.H., and Isaac S. Kohane, M.D., Ph.D. N Engl J Med 2012; 366:2240-2242 (Different Dr. Mandel) which builds this case...  although I think underplays some of the issues of connectivity and integration.

Here's a conversation I had with Kohane a while back, note the part in the middle where I talk about universal health URI architecture...

And finally, maybe we should jettison the fantasy notion that we are creating a single, monolithic EHR covering the full range from psychiatric data, neurosurgery details, exercise logs, and billing information.  Maybe things are just too amorphous, customized, and constantly changing to call this a single "record."  Let's just call it a "space" and get started with it.

 

SMART, VistA, and the Semantic Web

conor dowling's picture

yes to "space", no to "record" as you say - the latter leads to emailing
data dumps. Yes to patient data as secure but still just data that should
share the same space as other health-data - knowhow, institution
descriptions. I think the web of data provides that space.

From the paper ...

"Loss of technological leadership reflects apathy and even opposition
by EHR vendors to promoting *liquidity of the data they collect*. This
attitude has thwarted medicine's decades-long quest for an electronic
information infrastructure capable of providing a dynamic and longitudinal
view of the health care of individuals and populations"

Frankly this is a problem, a big one, for VistA too. For 30 years, the VA
and its consultants have coded custom, narrow bore, overlapping FileMan
data extractors, some that make HL7, some email and now XML. No
comprehensive approach to data management was taken, despite the facilities
invested in FileMan. And this continues, right now - look at the VPR
extractor, more custom code for custom data retrieval.

This waste of effort and opportunity is not about money - the pull of
"doing what you've always done" is as inhibiting as any support contract.

On Tue, Jun 26, 2012 at 11:11 PM, Tom Munnecke <munnecke@gmail.com> wrote:

> I might point out Mandel and Kohane's paper Escaping the EHR Trap — The
> Future of Health IT <http://www.nejm.org/doi/full/10.1056/NEJMp1203102>Kenneth D. Mandl, M.D., M.P.H., and Isaac S. Kohane, M.D., Ph.D. N
> Engl J Med 2012; 366:2240-2242 (Different Dr. Mandel) which builds this
> case... although I think underplays some of the issues of connectivity and
> integration.
>
> Here's a conversation I had with Kohane<http://www.youtube.com/watch?v=VS8AIRFsxSw>a while back, note the part in the middle where I talk about universal
> health URI architecture...
>
> And finally, maybe we should jettison the fantasy notion that we are
> creating a single, monolithic EHR covering the full range from psychiatric
> data, neurosurgery details, exercise logs, and billing information. Maybe
> things are just too amorphous, customized, and constantly changing to call
> this a single "record." Let's just call it a "space" and get started with
> it.
>
>
> --
> Full post: http://www.osehra.org/document/smart-vista-and-semantic-web
> Manage my subscriptions:
> http://www.osehra.org/og_mailinglist/subscriptions
> Stop emails for this post:
> http://www.osehra.org/og_mailinglist/unsubscribe/806
>