search

OHDSI / CommonDataModel

Definition and DDLs for the OMOP Common Data Model (CDM)
https://ohdsi.github.io/CommonDataModel
903 stars 453 forks source link

Have you looked at mCode for the Oncology? #268

Closed mtomas7 closed 3 years ago

mtomas7 commented 5 years ago

I found this initiative interesting:

ASCO’s mCODE seeks to compose lingua franca for cancer informatics: https://cancerletter.com/articles/20190607_1/

Conversation with The Cancer Letter ASCO’s Miller: mCODE will improve downstream analytics with essential cancer data elements https://cancerletter.com/articles/20190607_2/

Perhaps it would be possible to collaborate with them?

“There are several reasons why we think that mCODE will thrive,” ASCO’s past president Monica Bertagnolli said to The Cancer Letter. “First, it is a user-driven data structure, and its first version was created by a diverse team of researchers and clinicians who wanted to use the results to drive progress. “Second, it is iterative. The first version, mCODE v0.9, includes 73 data elements that should be reported for every cancer patient. It is pretty straightforward to see how having these elements in a standard format can be mapped fairly easily, and that doing so will make a big difference in data usability across different institutions,” said Bertagnolli, who leads the Alliance for Clinical Trials in Oncology Foundation.

The mCODE specification has a total of 73 elements across six domains.

“We understand it’s not going to serve every use case, or even the majority of use cases in oncology, given the complexity of the field,” Robert Miller, medical director of ASCO’s CancerLinQ, said to The Cancer Letter. “But we’re looking at this as a foundational set of data elements that in our opinion should be in every electronic record in standard computable structured format.

“But that’s where we are right now, that’s available for use by anybody.”

mCODE is available as an open source download here https://mcodeinitiative.org/access-mcode/?utm_source=home&utm_medium=click&utm_campaign=mcode

The initiative is funded by ASCO, FDA, MITRE Corp., and the Alliance Foundation. MITRE Corp., a government contractor that operates federally-funded research and development centers, serves as the technical facilitator.

cgreich commented 5 years ago

@mtomas7:

Interesting catch. We have spent a good chunk of time (too much, actually) last year figuring out the model and where we take the controlled vocabularies from. The task was to extend the OMOP model to capture the type of information necessary for cancer research. Because there you need to go much deeper: More histology, topology, grading and staging attributes, tumor-specific attributes like biomarkers, etc. The treatments are also more complex than in the rest of medicine (maybe with the exception of things like HIV).

It seems this mCode thing is a mini-OMOP (vitals and comorbidities) plus some oncology-specific thing. But if you look at the data dictionary and the value sets they are limited to ICD10CM, SNOMED and LOINC and we know this is not sufficient. So, this mCode thing seems more like a light-weight model for registries, which is what CancerLinQ is.