OHDSI Home | Forums | Wiki | Github

Conversion of measurement data beyond laboratory values: conversion of echocardiography into OMOP-CDM

Hi all,

I’ve just converted data of echocardiography into OMOP-CDM.
The mapping table I used is available at GoogleDrive enter link description here
I didn’t convert all data from echocardiography, but specific data that have clinical importance (such as ejection fraction, E/A, etc.).

I want to convert other kinds of data such as pulmonary function test or ECG.
Is there anybody who has experience for these job?

1 Like

@SCYou
Do you have a file with the data you want to map?

Hi Seng,

I am forming a vocabulary to cover Electrocardiographic (ECG) data originating from pre-hospital (Emergency Medical Services) and in-hospital sources - from electrocardiographs, 12-lead capable defibrillators, and ECG management systems covering the top device and ECG management systems. The ECG vocabulary would cover mapping of ECG lead level measurements and unconfirmed/confirmed diagnosis findings to standard OMOP concepts where possible and original source vocabularies organized by individual vendor and/or programs.

The purpose is to provide OMOP support for medical device based cohort discovery to augment and supplement EHR based cohort discovery.

What are you thinking in this context?

  • MK

======================================================================
sSS
[===]D Manlik Kwong
Engineer/Scientist
Institute for Clinical Research and Health Policy Studies
Tufts Medical Center Boston MA

Senior IT Program Advisor
Tufts Clinical and Translational Science Institute
Tufts Medical Center Boston MA

E-mail: mkwong@mailto:mkwong2155@gmail.comtuftsmedicalcenter.org

======================================================================
NOTICE OF CONFIDENTIALITY
This email message, including any attachments, is for the sole use of the intended recipient(s)
and may contain confidential and privileged information. Any unauthorized review, use, disclosure
or distribution is prohibited. If you are not the intended recipient(s), please notify the sender
immediately by reply email or telephone and destroy all copies of the original message.

Thank you @aostropolets , @mkwong

My next target database is also ECG.
My institution open the free accessible ECG database (ECG-ViEW II, you can see the paper [here][1])

So, I need concept ids for ‘PR interval’, ‘RR interval’, "QRS interval’, ‘QT interval’, ‘QTc interval’, ‘P wave axis’, ‘QRS axis’, 'T wave axis.

Thank you again for your interest and comments!
[1]: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0176222

Hi Seng,

For lead measurements - at least the basic ones like “Global” and lead specific - there are LOINC based concepts in the standard OMOP vocabularies. Some measurements programs like those from Physio-Control have the basics, while those from Philips TC50/70 for example have more extensive global and lead measurements available - so you will need to have both a custom and “standard” mappings - at least that is how I’m handling this for GE12SL, Philips, and GLASGOW based electrocardiographs and defibrillators.

Then there is the concept mapping of the 12-lead Interpretation statements, modifiers, and qualifiers - again dominated by GE12SL, Philips 0B, and GLASGOW. On top of that - I assume you would be mostly interested in confirmed 12-lead ECGs (vs unconfirmed) - which means you will have place holders for over-read cardiologist free text entries, edits, and deletions if you really want to dig into cardiologist overread accuracy, or initial findings, etc.

I’m hoping to have a proposed vocabulary posted to OHDSI by the end of the year to cover the above commercial measurements and criteria - including my own. Let me know what ECG devices and management system you are using - I may already have an OMOP mapping and software worked out.

Best - MK

@mkwong, @SCYou and friends:

So, we have LOINC, we have @SCYou’s lists. What’s next? Should this be a set of conventions with preferred/predefined vocabulary entries like for smoking? Where are we going with this?

Hi,

We (@kchan) are trying to building a datamart at Northwestern that includes all of the major cardiovascular diagnostic testing. We are obtaining both structured data from our source systems and running NLP on notes for additional variables. We are starting with echocardiography, but are also planning to work on cardiac catheterization, ECG, cardiopulmonary exercise testing and many others.

We found a couple links described a data model for echo. We are wondering if there has been more work done on echo or other cardiovascular testing. And if there are any working groups we could join that are working on organizing cardiovascular data.

Thank you in advance for your time.

Sincerely,
Faraz and Kyle

Gents:

There have been a couple attempts to get EKGs organized: Here and here. I don’t think folks have actually finalized this. You have the opportunity to do exactly that, @Faraz15 and @kchan.

Happy to help.

Hi Faraz and Kyle - when you get to importing ECG (typical 12-lead from electrocardiographs) data I have created maps for electrocardiographs from current GE - Mac5500, Philips - PageWriter, and Glasgow ECG measurement and computerized interpretation statements (un-edited) to SNOMED and LOINC concepts. Happy to send you the mapping for the electrocardiograph you are using. Each manufacturer also include more ECG measurements than what I have found a mapping to - so these are added to our OMOP vocabulary/concept as a custom vocabulary for the particular vendor program/version.

  • MK

Thanks. Do you two have thoughts on the best way to move an effort forward around either ECG or more broadly around cardiovascular diagnostic testing (echo, EKG, cath to start). It sounds like @mkwong was planning for a submission for ratification.

Manlik, if you are able to send us the maps for each vendor, that’d be very helpful. I believe we have GE, but regardless, we’d like to compare them and also see how they align with the data model used in cardiovascular cohorts (such as the Minnesota Code Classification System).

Thanks again,
Kyle

Hi,

We’re trying to map Canadian (Ontario) echo data to some common vocabulary. Were there any finalized suggestions here? For instance, is the common practice to map echo data to devices where appropriate, and procedures where appropriate? Are LOINC codes the most accepted vocabulary?

Any suggestions as to where to start would be very helpful.

Many thanks,

Kieran

Hi - I’m mapping our veterinary (2D, Doppler, M-Mode, and M-Mode Normalized) and human (TTE/TTE Adult in 2D Dopper and M-Mode) echocardiograph data to our respective OMOP warehouses. I do create a procedure_occurrence records and then using LOINC for measurements. Not all echo measurements have been mapped as I could not find a good representation for some of these.

For M-Mode - I’ve mapped 25 measurements. For Doppler I found 28 measurements. The rest may go into a custom vocabulary in the short term. This is work in progress.

Hi @mkwong ,

Thanks for your feedback, it was very helpful.

Kieran

Hi @mkwong ,

One more question regarding the mapping. Did you ever have an instance where a echo cardiograph data was mapped both to measurements and procedures? Or was it always one or the other?

Many thanks,

Kieran

Hi @Kieran

It depends - most of my ETL will generate a procedure entry and a set of measurements/observations. If it is a device like an echo, electrocardiograph, or single use device - I will also generate a device_exposure entry as well. This way it supports queries based on measurements, a procedure, and devices involved in the patient care.

Hello!

Sorry if I am asking this question at the wrong place!

I wanted to ask in the following LOINC term

“LOINC 59111-5
Heart valve E velocity/Annulus E wave by US”

Which heart valve is the term referring to? Is it the mitral (left-sided) valves or tricuspid (right-sided) valves or both the valves combined (which unfortunately will not be very useful since I need valve specific data)

I am working on a project where we are trying to use data from LOINC and it would be very helpful if someone could clarify that for me!

Thank you so much! @SCYou would appreciate your input!

t