OHDSI Home | Forums | Wiki | Github

Collaborating and coordinating methods research

(Martijn Schuemie) #1

A huge amount of methods research is taking place in OHDSI, and I’m very happy that a lot of this research crosses institutional and geographical boundaries. But this begs the question: how best to collaborate and coordinate? Because I think that collaboration and coordination are the key to accelerating our combined research efforts. I truly believe the whole can be much greater than the sum of the parts.

We have the Patient-Level Prediction Workgroup and the Population-Level Estimation Workgroup, each having separate meetings once in a while, but I’m not entirely happy with those. Most of the time, at least in the PLE meetings, it has been me talking and presenting some half-baked idea, and although I get a lot of value out of that, it doesn’t meet the goal of collaboration and coordination. I think @Rijnbeek and @jennareps have similar experiences in PLP.

I find that bilateral meetings, just focused on a specific topic (e.g. a PhD student’s research) is much more effective in terms of getting work done. So I definitely want to have more of those, but these tend to be closed to the public, so not in line with our open science philosophy.

One idea I’d like to propose is to merge the PLP and PLE meetings, to have a single methods research meeting on a regular basis (still one per hemisphere). I would also like to ‘push’ more people to present their unfinished research in these meetings. For some reason, for ‘public’ meetings people are incredibly reluctant to present their ongoing work. Often when I ask someone to present at a PLE workgroup meeting, the response is “Maybe in 2 months, when the paper is done”. But I would argue we would all benefit from discussing research much earlier than that.

Sorry for the rant. Looking forward to hearing what everyone has to say, including @Daniel_Prieto, @Rijnbeek, @jennareps, @msuchard, @hripcsa, @chenyong1203, @Nicole_Pratt, @SCYou, @David_Madigan, @nigam.

(Seng Chan You) #2

@schuemie I cannot agree with you more. And I can push Korean researchers to present their works if we merged those two WG meetings.

(Daniel Prieto-Alhambra) #3

dear @schuemie and all,
you have a point, and I also think we could do better if we used these fora and tc/webinars to discuss more preliminary work. I will discuss with my group and ask my PhD students to start presenting in these meetings.
Happy to help set up tcs etc if at all useful


(George Hripcsak) #4

If the leaders in the two WGs think it is best to merge, then that is fine by me. As we shift to patient-level causal estimation, it makes sense anyway. How do these relate to LEGEND meetings? Has LEGEND displaced PLE?

A separate question is engagement. Definitely be good to hear earlier research. Just thinking, if someone has new research to present, then another proper place may be the Tuesday noon (ET) community call so that everyone can hear, and that schedule is not full.

(David Madigan) #5

Great idea! These areas are converging and the silos are not helpful. Getting people to talk/present remains a problem.

(Yong Chen) #6

@schuemie I agree with you entirely. I believe it is a good idea, and there could be synergistic values generated from merging the PLP and PLE groups. I think the interactions among the merged presenters and the audiences from two groups can be interesting. I will be happy to present our work in one of the future meetings. We can discuss about the schedule to fix a specific time.

(Peter Rijnbeek) #7

Hi Martijn,

I agree. There have been merging efforts already, e.g. the Heterogeneity of Treatment Effect of @arekkas that combines prediction and causal inference in one OHDSI R Package.

The last TCs we had PhD students present on the PLP meetings which worked well. Unfortunately, these were very mostly my students and I would like to see others present more often. I think it is good to present unfinished work to get input from the team.

I do think we need to think about a strategy for authorship in this context. My preference would be to collaborate as much as possiible and avoid competing teams in the OHDSI community.

(Martijn Schuemie) #8

Responding to @hripcsa:

I think LEGEND meetings are separate. By its own rules, LEGEND doesn’t do method development, and these meetings would be all about method development.

Agreed we should also communicate exciting ongoing research at the community calls. However, that is a very different (more generic) target audience.