loinc, extending indivo data model for glucose results

[Benjamin West]

So I took a look at Indivo/SMART again, again.

http://docs.indivohealth.org/en/v2.0.0/coding-systems.html

http://wiki.chip.org/indivo/index.php/HOWTO:_install_Indivo_X

http://dev.smartplatforms.org/reference/rest_api/

http://dev.smartplatforms.org/reference/data_model/

http://dev.smartplatforms.org/reference/app_manifest/

(@I went over to LOINC and downloaded the codes.  Grepping through the files indicates over 300 codes for glucose.

Maybe we can get a pharmacist or insurance adjuster or someone to help identify the right codes.  I anticipate using about a dozen or so?

I took a stab at trying to identify a single value to encode/query Indivo: a finger stick glucose test.  I documented my attempt here: https://gist.github.com/bewest/4965993

I whittled down the candidate LOINC codes to about 20, but I still have no idea which one is the right one.

For that matter, Jana (~jebeck) has documented several types of "lab results?" from her CGM, and I have documented several results from my insulin pump for which I'm not sure any codes actually exist.

Is there any advice on minting new codes for these purposes?

@medevice, I've cross-posted to the smart-app-developers list, in hopes they'll have a suggestion.

@smart-app-developers - My medevice/diabetes group would is considering extending your data model so that diabetics have a comprehensive set of tools.  Overall we like the basic data model.  I figured a simple fisrt step would be to represent a bunch of glucose finger sticks using your data model, but finding relevant codes seems to be a limiting factor.  Should we simply make some up?

-bewest

[Benjamin West]

Josh,

Thanks very much.

I'll give the search tool a shot.

I'll follow up in a bit with an updated gist, but wanted to respond to 2:

This is my intuition as well.

I think what makes most sense for me, in putting all this together, is to have some kind of git-phr tool that creates and stores a serialized version of a single patient's ehr graph in a git repo.  This helps me as the user retain control over who has made what changes to my medical record.  I wonder if this makes sense to anyone on the smart/indivo side?  Eg, serialize my graph, and use git to do audit tracking and establish transport independence?  This way, all I have to do during my device check-in process is add data to my raw (text?/n3?/json?) records, update the manifest, and git commit; git push.  I understand you are doing a per-record level type of write transaction using POST/PUT, but unclear how audit trails are captured.

Otherwise we are currently undertaking exactly the kind of inventory you suggest, will have more soon, I hope.  I thought the finger sticks would be a nice entry point, it appears if I look hard enough many of our values will be there.  I see that there are even codes for estimates and averages.

You may find the csv under the analysis/bewest-pump directory more useful.  Some csv is from a test pump with little data.  I updated the gist comments to help reflect this.  I have additional data I'd like to upload soon; at some point I imagine someone will update our wiki or pages with this material.

Thanks very much!

HTH,

-bewest

On Sat, Feb 16, 2013 at 11:16 AM, Josh Mandel <jma...@gmail.com> wrote:

Hi Ben,

Great work already, just pulling resources together!  A couple of quick thoughts on LOINC codes and modeling:

1.  You may have an easier time making sense of LOINC using either the online search tool (http://search.loinc.org) or the downloadable program RELMA.

2.  My initial instinct (which you may disagree with) would be to explicitly separate out "medical" data into glucose readings and insulin administration snippets; and then separately model the "device" data (battery status notifications; reservoir state, etc).

3.  For finger sticks (measured by glucometer; units = mg/dL?), you should probably be looking at LOINC code 41653-7 http://s.details.loinc.org/LOINC/41653-7.html .

And a question: looking through the following log, I couldn't easily identify the actual glucose readings: https://github.com/bewest/decoding-carelink/blob/rewriting/Carelink-208850-2006-07-export.csv

Where are they?  

Best,

  Josh

--
You received this message because you are subscribed to the Google Groups "SMART App Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to smart-app-develo...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

--
You received this message because you are subscribed to the Google Groups "SMART App Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to smart-app-develo...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.