Alistair Johnson

This is likely a bug in NegBio - you can checkout my pull request and should find it works (I just tested it). See https://github.com/ncbi-nlp/NegBio/pull/20

Unfortunately we don't have the bandwidth to support a Windows ETL - but we'd welcome a pull request if you get one working. Essentially step 0 defines environment variables which...

@parisni is saying you can cross reference the itemid with the mapped concept because measurement_source_concept_id maps to the itemid and measurement_concept_id maps to the standardized identifier. As you guessed a...

There is no specific "measurement" concept_id - all columns which end in `concept_id` reference the same CONCEPT table. That being said, the data present in the `measurement_source_concept_id` column have a...

I should add that most standard ontologies (SNOMED, LOINC, etc) already have `concept_id` assigned, and that etl only creates new `concept_id` for the local codes present in MIMIC-III.

Hm well I think we can rule out drug_era and dose_era because (1) they are derived and this is real data and (2) they explicitly state in the documentation that...

Updated function here: https://github.com/MIT-LCP/mimic-omop/commit/e1f5b984f2a0b120591aa2eb09fe7514453d5bc7 Both extracts use `looks_like_value` to check if it's a valid number (no matter the format). Also did a fix to pick up the number `1,000,000` here...

Okay, let me check!

Hm I see no differences but the ETL seems broken for negative lab values - it drops the negative.

Yeah not very well documented apologies. I am pretty sure the default concepts which are checked for download on the website were the only ones used. I successfully ran the...