Closed vojtechhuser closed 6 years ago
Yes - the makefile is quite simple though. It calls a single SQL script: make-concepts.sql
. This script creates all the subsequent tables by calling all the various SQL files.
PostgreSQL (or "standard" SQL if you prefer) is the desired syntax. I think creating an individual folder like you suggest (OMOP) would work well.
We have another colleague looking into OMOP and its suitability for MIMIC-III. @parisni any comment?
Thanks for the comment. It would be great to have more people in the mimic-omop team. awaiting @parisni reply.....
Hi @vojtechhuser I plan to work on that topic too. I would be glad to collaborate. My goal is to have something working before 2018
Just an update - I just drafted PERSON table ETL. I am working with @parisni on this. Anyone is welcome to chip in.
at OMOP symposium, number of other folks with mimicIII data in OMOP format were consulted. Please use this thread to report any outcomes.
This group from Georgia Tech also has m3 mapping - See here: https://github.com/gt-health/omop_etl_public/blob/master/dataset_mappings/mimic3/mimic_stage_insert.sql
The mapping of MIMIC to OMOP is being carried out in a separate repository: https://github.com/MIT-LCP/mimic-omop/
Contributions welcome!
Looks good. Any ideas on sizing? The mimic-code project says I should give my Postgres about 100G providing room for the mimic tables, indexes and etc. I guess I need to add to that for the OMOP tables. Are they going to be bigger or smaller than the raw mimic tables?
I would like to contribute to concept folder a view of the data that emulates some of the tables in the OMOP CDM.
Do all concept folder contribution get build automatically?
Is the syntax for any contributed view Postgres SQL flavor?
Do I just submit PR to the repo with concept/OMOP/person.sql code? (if I want to start with person table?)
per readme file
related link : http://forums.ohdsi.org/t/mimic2-and-possibly-mimic3-data-in-ohdsi-cdm-format/860