OHDSI Home | Forums | Wiki | Github

EHR Dream Challenge: Patient Mortality

General
#62

Hi! My name is Artem. I’m with Vocabulary team here at OHDSI. I’m an anesthesiologist by profession and have a strong interest in data science also. I’ve found about this challenge recently and would like to join the team. I’m aware that from today round 1 of leaderboard phase starts and you have already covered a lot of tasks. But still I think that this would be great experience and I am happy to contribute.

#63

@alexander @Vojtech_Huser @rohitv @Chungsoo_Kim et al: Take a look at this thread https://www.synapse.org/#!Synapse:syn18405991/discussion/threadId=6145

Gives us more info on submitting.

RE: a call this week – if folks want to chat I can put a WebEx together. I am traveling today and can’t join myself.

#65

I got this letter yesterday:

Hello @EHRChallengeParticipants,

We are pleased to announce that the Leaderboard Phase, Round 1, is officially open.

You can submit your models to the “EHR DREAM Challenge Submission” queue to be officially scored against UW data. If you want to test your models without running against the UW data, the “EHR DREAM Challenge Fast Lane” will continue to be available.

You will be allowed to have three successful submissions per round (9 total over the course of the challenge). If you submit a model and it breaks, that will not count toward your three.

In the UW environment, your resources will be.
70 Gb RAM
4 Cores
6 Hours runtime

Your submissions and their progress and scores will appear in this dashboard

If you have questions or if the returned log files are vague, leave a comment in the discussion board and we will help debug your models.

Good luck!
EHR DREAM Challenge Admin Team

So we have 6 Hours runtime now. But they limited number of submissions. Please, be careful when you submit anything there. We can do it only 3 times…

2 Likes
#66

Hi guys. I’ve spent a few hours trying to create and connect to the Postgres database inside of the docker container and something goes wrong. Can someone else try to do it? I tried a few approaches to do it, but I’m not sure which is the right one.

#67

Hi all.

It was pretty busy last week and I will try hard this week instead. I’m creating a plp package now. When the cohort is complete, I’ll put it in a package and build some models and run them. Still, the docker is a difficult part for me. I will learn how to release it.

Also, I’ve put some comments in the word document about creating a cohort.

Thanks
Chungsoo Kim

2 Likes
t