OHDSI Home | Forums | Wiki | Github

Looking for phenotype definitions(in EHR) for Diabetes and Depression


(Abhi) #1

Hi Folks

Despite lurking around on OHDSI forums, I am still a newbie here and have a naive question. We have a project where we need to pull up patient records from EHR that matches broadly to two disease categories.

  1. Broadly - “Diabetes/Obesity”
  2. Depression

I am sure there are groups that have previously mined EHR data corresponding to these health conditions. I am curious to know what cohort selection criteria researchers have used to define phenotypes for these two conditions. Also is there a place where OHSDI catalogs such phenotype definitions that could be re-used by others.

Thanks much!
-Abhi


(Anna Ostropolets) #2

OHDSI catalogue with phenotype definitions would be cool indeed :slight_smile:

The most up-to-date definitions can be found here: https://atlas.ohdsi.org/#/cohortdefinitions
You can look up you conditions of interest by simply typing the names in the cohort definition search tab. Don’t be intimidated by COVID prefix - they were populated for COVID hackathon but are in fact COVID-agnostic. Some of them were also used in OHDSI papers such as this


(Abhi) #3

Thanks @aostropolets


(Andrew S. Kanter, MD MPH FACMI FAMIA) #4

Thank you @aostropolets! It occurred to me that it would be nice to be able to have a discussion about a specific cohort definition or concept set. Does Atlas allow that? For example, looking at the prevalent type 2 diabetes cohort and the relevant concept set, I see all sorts of included concepts that don’t specifically identify which type of diabetes. Seems like a concept set author would like to be able to get feedback.


(Anna Ostropolets) #5

That’s a very good point. We have a project in mind that relates to storing such phenotypes outside of Atlas. There you’ll be able to create a new phenotype if you feel that the existing one doesn’t meet you criteria. But I think that having a platform (or maybe just ohdsi forum thread) where you can actually discuss the phenotypes would be beneficial indeed. Tagging @Gowtham_Rao and @Patrick_Ryan


t