OHDSI Home | Forums | Wiki | Github

A Unified Generation

Today we have different ‘Generate’ experiences across multiple use cases within ATLAS. I would like to propose that we develop a shared component to provide a simpler and consistent user experience. To begin I thought it best to enumerate all the particular ‘Generate’ use cases we have today - please let me know if I’m missing any.

  1. Generate a Cohort Definition ( or use a cohort definition to write records to the cohort table in the results schema )

  2. Generate a Report ( within cohort characterization, currently on the reporting tab within a cohort )

  3. Generate Evidence ( for negative control exploration, from the explore evidence tab on the concept set page )

  4. Generate Incidence Rates ( from the incidence rates drop down or generation tab )

  1. Generate Population Level Effect Estimation ( from the results tab of the population level effect estimation page )

Once we determine that we have a complete list we can iterate on possible designs for a consistent component. We should also keep in mind some future use cases that are on the roadmap such as generating Patient Level Prediction Models.

@Frank Excellent. I am not aware of any other.

The #1 is missing a screen, but it looks a lot like the one in #3. Tho, not sure what ‘distinct people’ means in #3 in that case.

The topic here is a unified generation, but the cases above have both generation and results reviews co-mingled together, possibly confusing what we’re trying to focus on: I’m assuming that we’re talking about a unified experience for selection of sources to execute analysis against? I can imagine that being unified, but when we have the results (which different analysis may have different summary results to render) it complicates things.

Once we get into the discussion of our options, I will have some things to say about how putting the generation co-mingled with results has had negative impact on the application.

t