OHDSI Home | Forums | Wiki | Github

FAIR hackathon at Bio IT World - OHDSI topic?

Hello everyone,

On May 14 and 15, CHI will organize a FAIR Data Hackathon in conjunction with the Bio IT World conference in Boston. I was wondering if there is any interest to make some of the OHDSI data sources or infrastructure FAIR as part of this hackathon? For example, we could work on metadata descriptors for some of the synthetic datasets used to test OHDSI, or any open datasets that we can find and use (NHANES maybe?) in OMOP format, and work on improving the FAIR ranking of those datasets. Let me know if there is any interest and I can propose this topic to CHI!

(If you’re not sure what FAIR stands for, the 15 FAIR Data principles are explained here: https://www.dtls.nl/fair-data/fair-principles-explained/)

Greetings,

Kees van Bochove, The Hyve

I’m not currently planning on attending Bio IT but would happily attend a hackathon. :smile:

@Frank, you’re always welcome to visit Boston!

That said, BioIT is 2 weeks after the OHDSI F2F meeting in NYC. We may find that it would be difficult to get enough momentum to give this the support it needs. @keesvanbochove as we saw last year, there aren’t a lot of OHDSI collaborators who attend BioIT World primarily because it does tend to be more pre-clinical/clinical R&D focused. However, in a world where we are starting to play more with i2b2/tranSMART and the like, it may start to come up on more people’s radars.

Side bar: there’s a really nice Deep Learning in Healthcare conference in Boston the following week (May 24 & 25) that may also be of interest to people looking to collocate in the region around crunchy questions that impact us.

This is all to say… you guys are all more than welcome in Boston! Please come!

the link requires password. is there a better link?

Can you explain a bit more how FAIR score is computed. I am very interested in advancing metadata.

As far as I can understand there is no unique FAIRness score (yet).

FAIR metrics group (http://fairmetrics.org/) has recently published a paper referring to a process of defining “FAIRness” metrics. However, they define metrics for each independent FAIR principle per se, and not an overall score of FAIRness.

Thanks @Frank and @krfeeney for your responses! If the timing doesn’t work out this year, we can mark it as an idea for Bio IT World 2019, if the FAIR Hackathon will still continue then. Or see if we can organize it around the OHDSI meeting in October.

But if there is more interest it would be great if we could do this. I’ll expect we will be working quite a bit in the coming years on increasing the FAIRness of OHDSI tooling, and getting input from the community now would be helpful to prioritize that work and understand requirements and possible applications.

@Vojtech_Huser that’s odd, it should not ask for a password and doesn’t if I retrieve the website now. In any case, the original article in Nature Scientific Data with the principles is here: www.nature.com/articles/sdata201618
If you want a deep dive into a possible way to realize all principles without imposing any APIs or data models using semantic web technology, have a look at Interoperability and FAIRness through a novel combination of Web technologies [PeerJ]. I don’t think we should start with that approach, because there are a number of much easier steps we can take in OHDSI towards FAIR (starting e.g. by assigning permanent identifiers to mappings, putting basic metadata, versioning the vocabularies etc.) but ultimately, once we have the right supporting tools, this would be a very nifty way to do ETL at scale, reproducible, and as a community effort (as opposed to one dataset at a time). Instead of mapping the source data to OMOP, we would map OMOP to the data :slight_smile: :smile:

That being said, it’s good to realize that mapping to OMOP is actually a great step you can take today to make medical history data more FAIR, with emphasis of course on the I of Interoperability.

1 Like
t