Reproducibility In The OHDSI Environment — Why It Matters, And How We Improve Upon It
The March 29 OHDSI Community Call focused on reproducibility, and three leaders in the community shared presentations on this topic.
– The 2021 OHDSI Reproducibility Challenge (Anna Ostropolets) – Creating Reproducible Studies (Martijn Schuemie) – The OHDSI Reproducibility Service (Asieh Golozar)
The videos and slides from each presentation are available below.
Anna Ostropolets, a PhD student at Columbia University, shared a presentation on the OHDSI Reproducibility Challenge, a one-day event held during the 2021 OHDSI Symposium that attempted to have teams of OHDSI collaborators reproduce the cohort logic for target, comparator and outcome cohorts for a published study run outside of OHDSI.
Martijn Schuemie, a Research Fellow in epidemiology analytics at Janssen Research and Development, shared a presentation on creating reproducible studies using OHDSI tools and best practices. Topics included the need for analyses to be repeatable, writing studies as a pipeline, and how to create the proper study package.
Asieh Golozar, Vice President and Global Head of Data Science at Odysseus Data Services, Inc., shared a community update on the OHDSI Reproducibility Service, which is centered at The Roux Institute at Northeastern University. This talk includes discussion of what type of study makes for a good candidate for reproducibility, the approach with the Center, and a proof of concept study that is seeking data partner collaboration.
The 2021 OHDSI Reproducibility Challenge (Anna Ostropolets)