All of US EHR is under OMOP/CDM format. I wonder if tools, e.g., Generating Features for a Cohort • FeatureExtraction, can be used in its cloud computing environment?
I heard it’s hard with their env. They primarily use Jupyter notebook.
Yes, you are correct. Makes hard things easier but makes simple things complex. If one is a coder, it is a net negative.
Our group at Northeastern uses AllofUs quite a bit and we have not been successful with any of the OHDSI HADES packages. Primarily the limitation as far as I can tell is the read-only access to the db - there is no write schema / scratchpad which is often required by OHDSI packages. There are some limitations on temp tables as well.
AllofUs also does not follow some OHDSI conventions for some tables (e.g., the observation period table is not implemented correctly - some participants have > 1000 observation periods…) which also makes it difficulty to use some of the OHDSI tools right out of the box.
I agree that it is not optimized for it.
At some point, we had r4aou set of scripts.
New page is https://github.com/cmayer2/r4aou
old page was https://github.com/lhncbc/CRI/tree/master/AoU
As an update for anyone stumbling on this thread later. @louisahsmith and I have released an R package that will hopefully be useful for OHDSI folks who’d like to use the AllofUs data (Interface for All of Us Researcher Workbench • allofus). Hopefully we will be able to continue to add compatibility with OHDSI tools over time. Thanks @Vojtech_Huser for those links - some of the code was very helpful for us to include a function to import ATLAS cohort definitions.
1 Like