THEMIS: New Initiative to create definitive conventions for data in the OMOP CDM

SuMan_Nam · September 28, 2017, 4:30am

I’d like to join, too.
My email is sumannam@gmail.com

bobmalan · October 24, 2017, 2:21pm

Count me in as well.

Vojtech_Huser · October 25, 2017, 9:02pm

Inspiration for the group: (draft v4 of pcornet)

Christian_Reich · October 26, 2017, 1:13am

@Vojtech_Huser:

Do you have the full thing? Could be good for the THEMIS discussions. Even though some of the choices are debatable.

jweave17 · October 31, 2017, 4:41pm

Hi @Christian_Reich, could you please add me to this group? I’d like to join the discussion on how to best and consistently capture pregnancy episodes and mother-child linkages, among things.

MPhilofsky · November 3, 2017, 6:33pm

@Asha_Mahesh @Christian_Reich

Are we keeping a list of issues that need to be resolved in the THEMIS group? I have some items to add

Melanie

Christian_Reich · November 3, 2017, 7:14pm

@MPhilofsky:

What do you have? The list is spread over the 4 groups at http://www.ohdsi.org/web/wiki/doku.php?id=projects:workgroups:themis.

MPhilofsky · November 6, 2017, 7:01pm

I am mapping source EHR social history concepts to OMOP. Some of the social history concepts (i.e. alcohol, tobacco, illicit drug use, sexual activity) are in the Observation & Measurement domain, some are Condition_Occurrences, but aren’t true “conditions”. We need conventions on which standard concepts to use, more standard concepts to represent source values and maybe some “suggested” concept sets. There are many standard concepts to choose from to represent the fact that a person drinks alcohol. And no standard concepts to represent their denial of illicit drug use.

Also, I am using the Fact_Relationship table to link two Observation_ids together or a Condition_Occurrence_id to an Observation_id. However, there are only 10 standard Relationship concepts and none are appropriate.

I’m sure I’ll come up with more examples as I work through our social history source data.

SCYou · November 7, 2017, 12:15am

Standardization of observation from survey data or social history in EHR is what I needed for a long time. In Korea, many hospitals want to convert their health examination data into CDM, too (@whk) . Main obstacle for this is mapping social and medical history.
So, I absolutely agree with @MPhilofsky 's idea!

And I want to add some of my opinions or questions about observation mapping:

The dose unit for amount of alcohol should be standardized, too
In Korea, many survey asks ‘how many cup of ‘soju’ do you drink?’. ‘soju’ is Korean local alcohol. So, I converted the dose unit for alcohol history into ‘unit of alcohol’. I think the dose unit for alcohol history about wines, beers and whiskeys should be standardized for future research, too.
The concept_id I used
I would be not easy to read, but I mapped some social and medical history of Korean national survey data(here).
The below is the concept_ids I used for social history:

Tabacoo
4276526 Cigarette smoker
4222303 Non-smoker
4310250 Ex-smoker
(I couldn’t find ‘current smoker’)

40766929 How many cigarettes do you smoke per day now [PhenX]
40766930 On the average of the entire time you smoked, how many cigarettes did you smoke per day [PhenX]
40766364 How many cigarettes per day do, or did, you smoke [PhenX]

Alcohol
40771103 How often do you have a drink containing alcohol [SAMHSA]
3037705 Alcoholic drinks per drinking day - Reported

Exercise
4036426 Frequency of exercise

2.history of previous medical condition vs. family history:
the current SNOMED concept_id for family history has both concept_ids for ‘No FH’ and ‘FH’
(eg, concept_id=4053372, concept_name= No FH: Hypertension
vs
concept_id=4050816, concept_name = FH: Hypertension)

But it is not easy to find ‘has no history of medical history’ for previous medical conditions
eg, Is there any concept_id for ‘no history of cerebrovascular accident?’( concept_id =4077982, concept_name= History of cerebrovascular accident)

So I applied different strategy for mapping medical history(only existence of medical history is stored) and family history(existence and non-existence of family history are stored)? I think 'without history’is also quite important information. I think this should be standardized, too!

Dymshyts · November 8, 2017, 4:03pm

Hello:)

I’m working on SNOMED concepts domain definition right now.

So please, share you ideas what might be changed:

which concepts have the wrong domain according to your observations?
did you notice the changes between, say, some spring release and the autumn one?
here was the big change Proposal to change the definition of measurement domain to avoid ambiguity
all the tests findings turned to be Condition instead of Measurement
do you have any CDM data representation problems, for example, some entity has Values or results, but you can’t repesent them as it’s a Condition or Procedure?
and any other related thoughts and ideas

Vojtech_Huser · November 22, 2017, 2:55pm

Yesterday, the measurement subgroup of Themis group had a phone call. Some people volunteered for various iteams on our todo list.
I volunteered for measurement concepts and units. If your site would like to help with the effort, we hope to base it on real world data. See the beta network study at this link:

github.com

OHDSI/StudyProtocolSandbox/blob/master/themis/extras/notes.md

Themis group (sub-group: measurements) aims to standardize lab tests and related issues (e.g., units)

# Introduction
This effort will be based on real world data at OHDSI sites.

## Update 2018-05

Slight update to the data collected for ThemisMeasurement Database (looks at Measurements and Observations). Achilles Heel implementation discussion. Posting of units results.

## Update 2018-02

The result of the network study is ThemisUnits database (or knowledge base) (and other knowledge bases also looking at measurements (without regard to units)
JAMIA Open allows publishing short article centered about database. Outcomes will be published under this framework (as ThemisUnits database)  

## Update 2018-01
Partial interim themis study results can be found at https://github.com/OHDSI/StudyProtocolSandbox/tree/master/themis/extras/partial_results


# Execution details
Consider installing Arachne and run and submit the results in Arachne (preferred)

This file has been truncated. show original

SCYou · November 23, 2017, 10:48am

I appreciate your contribution, @Vojtech_Huser

In Korea, as growing number of hospitals is joining the OHDSI, we’re trying to make some ‘standard measurement set’. I expected we can make a set of about 300 measurements, which have clinical or academic importance. And then we will unify the concept_id and unit in Korea OHDSI network for this measurement set.

How do you think about this idea?

Furthermore, I’m converting echocardiography data into OMOP-CDM (to measurement table, of course). I’ll share the concept_ids I used in this work.

clairblacketer · December 7, 2017, 4:17pm

Hi,

Just wondering what the meeting plan is moving forward? I saw discussion of another F2F meeting in California after the new year but I didn’t see if the date had been finalized. Are we only doing F2F meetings for the large group and then meeting in the focus groups separately? I saw that focus group #3 had a meeting yesterday but it doesn’t look like the other focus groups have met yet. Are there plans for the other groups to meet as well?

Thanks!
Clair

Ajit_Londhe · December 11, 2017, 10:46pm

Hi Clair,

We’ve been doing bi-weekly calls, split up by sub-group: http://www.ohdsi.org/web/wiki/doku.php?id=projects:workgroups:themis. I’m not sure if the meetings are being posted on the OHDSI wiki, but here’s what I do know:

For Group 1, @mvanzandt has been leading it. We are voting on the next call, which should be sometime between Dec 19 and 21.

For Group 3, @Asha_Mahesh has been leading it.

@Asha_Mahesh, Minnie, @Christian_Reich, and @gregk – could the meeting URLs and dates/times be added to the above link?

Thanks,
Ajit

Gowtham_Rao · December 15, 2017, 10:24pm

Where is the information about when these meetings are held, and how to participate?

Christian_Reich · December 19, 2017, 10:16am

@Gowtham_Rao:

Here: http://www.ohdsi.org/web/wiki/doku.php?id=projects:workgroups:themis. Will add all that information shortly.

rtmill · January 2, 2018, 2:41pm

Putting that information on the wiki would be extremely helpful. In the meantime could we post the meeting information here so folks that want to contribute don’t miss the next round of meetings?

kDarko · February 27, 2018, 5:34pm

I just recently experience many of the pains outlined. How can I help? how can I subscribe to this working group?

Christian_Reich · February 27, 2018, 6:06pm

@kDarko:

Put your name into one of hte Focus Groups, and start attending their meetings. There is also a face-to-face coming up 8/9-Mar in 1000 Oaks, CA.

DTorok · March 15, 2018, 5:07pm

Where I can find the meeting schedules and call in information for all the Themis groups?