Blog Posts by Sam Hume

Results from the COSA Dataset-JSON Viewer Hackathon

We just wrapped up our 3rd Dataset-JSON COSA Hackathon, and this one may have been the best so far. All 3 Dataset-JSON hackathons have produced impactful outcomes and filled a need within the Dataset-JSON community. Since we started the PHUSE-CDISC-FDA Dataset-JSON pilot, folks have been asking me about the availability of Dataset-JSON Viewers. I’m thankful that I now have a good answer to that question.

Each of the 3 Dataset-JSON hackathons was organized differently. The first hackathon teams worked on any Dataset-JSON tooling that interested them. We stated a priority need for conversion tools, but there were no restrictions on the type of Dataset-JSON tools developed. That hackathon produced something like 21 projects and several conversion tools. Those same conversion tools allowed us to quickly start and successfully complete the PHUSE-CDISC-FDA Dataset-JSON pilot project. The second hackathon was a collaborative effort to produce a draft Dataset-JSON REST API specification. That API specification is now being used as a draft standard, and we plan to publish the final version later this year. Our third and most recent Dataset-JSON hackathon was our first competitive one. For this hackathon, teams were tasked with developing a Dataset-JSON Viewer. Several participating developers confessed that the competition motivated them to invest more time and effort into their projects, and the results were evident in the high quality of the work produced.

The Viewers created during the hackathon immediately fill a need within the CDISC community. Some organizations are testing Dataset-JSON for future use or implementing support for it in their products and have requested Viewers to function as part of their tooling. CDISC 360i will use Dataset-JSON, and the Viewers will be part of the 360i open-source tool set. Students attending CDISC’s new Dataset-JSON training class will benefit from Viewers to explore Dataset-JSON datasets. Overall, requests for a Dataset-JSON Viewer have come from numerous sources, including sponsors, regulators, vendors, and CROs.

When the Dataset-JSON Viewer Hackathon judges got together to compare our scoring and select winners, something unexpected happened - all 7 judges had ranked the top few projects the same. That made things easy. As a judge, I can say that all 7 Viewer applications submitted were solid. They did the basics quite well and were surprisingly capable given the limited time the developers had to work on them.

The winning Viewer by Dmitry Kolosov, called the VDE Dataset Viewer, does many things well. It runs locally on your laptop. It is easy to install, intuitive enough to start using without reading the documentation, has a polished interface, has some advanced filtering capabilities, and handled our very large dataset test well. Remarkable for a product developed during a short hackathon.

I believe that multiple Viewers may become widely used. Each project did some things well may satisfy different user needs. For example, the runner-up in the hackathon created a very nice web-based system that could be made available from a server. Another team built an R-Shiny application that may be popular in R shops. Others have a rich feature set that goes beyond simply viewing datasets, such as the Smart Submission Dataset Viewer with its conformance checking capabilities.

If you’re interested in seeing a demo of these Viewers, make sure to sign up for the next couple quarterly CDISC COSA Spotlight Webinars as the project teams will be demonstrating their Viewers. The Q1 COSA Spotlight on Tuesday, March 25th at 11 AM-12:30 PM ET will feature demonstrations from the hackathon winner and runner-up. I think you’ll be impressed by what they were able to accomplish in a short period of time.

Here’s where you can find more information on the Dataset-JSON Viewer Hackathon projects (in no particular order):

  1. VDE Dataset Viewer (Hackathon Winner)
  2. Open Dataset-JSON (Hackathon Runner-up)
  3. Smart Submission Dataset Viewer (Hackathon Honorable Mention)
  4. Dataset-JSON Viewer - R-Shiny (Hackathon Honorable Mention)
  5. JSON Visualizer (Hackathon Honorable Mention)
  6. EDA Dataset-JSON Viewer (Hackathon Honorable Mention)
  7. Dataset-JSON Visualization (Not an open-source project)

Although the Dataset-JSON Viewer Hackathon was competitive, one of the cool dynamics was the free exchange of ideas and information between the participants. There is no doubt that the open-source nature of these projects contributed to the exchange of ideas, and I hope this cooperative dynamic continues as the developers evolve their tools.

COSA hackathons promote the development and application of open-source tools that support the implementation of one or more CDISC standards. As the adoption of open-source continues to gain traction in our industry, hackathons have proven to be a great way to tap into that open-source energy and address gaps in our tool or feature set. As CDISC kicks off 360i, we anticipate hosting 360i hackathons to fill gaps in our automated end-to-end standards-driven research data pipeline.