OHDSI Home | Forums | Wiki | Github

Software Roadmap for Version 6


(Edward Palmer) #1

I’m confused around the software roadmap. Vocabularies, Achilles, and Atlas are all supported for version 5x only, and the milestones for reaching version 6 seem to be somewhat neglected.

Is there some hope for version 6, or should we be reverting to version 5?
This is particularly confusing, as the master branch of the CDM points to version 6, but all other indications would suggest that version 6 isn’t ready for implementation.


(Gregory Klebanov) #2

Hi Ed,

The current plans are to provide version OMOP CDM v6 in ATLAS v3. The ATLAS (@Patrick_Ryan as a product owner) team is currently working on ATLAS 2.8 which - at least theoretically - should be the last major release in version 2. The team is planning to start ATLAS 3.0 work sometime in early Q2, 2020 but it will take a few months for the release to happen

So, my personal view is that if you do need a mature working environment across all methods and tools - you better stick with CDM v5.3.1. That was my guidence to our commercial customers as well.

Btw, are there any specific use cases you are targetting that warrant CDM v6?

Also, I think what you are highlighting is a current gap that I have been also advocated to fix for a while - it is to have a single OHDSI roadmap across all key components - CDM, vocabs, methods and tools (ATLAS, ACHILLES). We have attempted something like this last year. But it ended up being not that simple since OHDSI is not an organization - it is an open science community that lives on contributions and funding of its community members and thus it is a bit difficult to commit to specific timelines. However, I think we should still revisit the idea of having a single integrated roadmap, even if it is more of a vision


(Edward Palmer) #3

Many thanks for the response. There are some nice features hopefully going to be implemented in version 6 that would allow us to better represent drug infusions for example. We don’t necessarily need robust tools, as we are comfortable debugging. More importantly, there is also a large change cost, as we are proposing everyone in our research network (7 major nodes at present) switch to an OMOP format. So whatever we do now, will likely be semi-permanent. I’ve just finished a reasonably large piece of work to parse a standard UK ICU data format into version 6. It would be a mild nuisance for me to re-write it for version 5, especially if that is going to be deprecated in the near future!

Hence we were really keen to implement version 6. Is there an open contribution method for Achilles/Atlas? If there are active development branches for version 6, it is possible that we could find some resource here to help contribute toward your version 6 milestones.

More broadly, from an end user perspective, a more joined up roadmap would be really appreciated! Version 6 has been canonical on master since Oct 2018, so it does give the impression that it is the version we should all be working on. Obviously immensely grateful for the hard work that goes into this.


(Edward Palmer) #4

After a long night of coding I’ve re-written our pipelines to CDM 5.3.1. It wasn’t as painful as I thought it would be. I think we can safely price in the cost of moving to 6 in the future. But some visionary milestones would be great. If they are clearly mapped out, then perhaps we can contribute to overlapping elements.


t