In 2020, multiple University Hospitals in Finland started initiatives to convert electronic health records (EHR) and registry data to the OMOP Common Data Model, thus enabling federated analysis of observational health data. The OMOP CDM and OHDSI tools were selected as part of a nationwide system to support the development of personalized treatment algorithms. One of the aims is to enable faster and more accurate cancer diagnosis and support treatment selection using evidence from data already collected by the hospitals.
The individual organisations are facing this challenge: to align on a common approach, develop shared processes and implement an overarching organisation to ensure that all partners share the same level of knowledge, expertise and understanding to agree on common conventions and best practices that facilitate the collaboration.
How we solved it
To address this challenge, The Hyve provided two workshop sessions of five hours each. The topics and agenda were set in consultation with key stakeholders prior to the workshop to ensure the topics were relevant to a diverse audience of medical experts and IT specialists.
During the workshop we provided insights in the following aspects of the OHDSI ecosystem:
- Current OMOP Common Data Model (OMOP CDM v5) including standardized vocabularies, structure and conventions
- New and ongoing CDM developments (OMOP CDM v6, Oncology module)
- OHDSI community resources, like the forum, The book of OHDSI, EHDEN Academy and code repositories.
- ETL building best practices and available open-source OHDSI tools (Usagi, White Rabbit, Rabbit in a Hat, delphyne)
- ETL validation best practices and available open-source OHDSI tools (Achilles, Data Quality Dashboard, CDM Inspection report)
- Step-by-step guide and demo of running a study with Atlas (cohort creation, inclusion/exclusion criteria, cohort export to RStudio to run R packages)
Since engagement and interactivity are key to a successful workshop, we included multiple Q&A sessions and a number of online polls. Besides, we collected use cases that were closely related to the clinical practice and research studies the participants experience on a day-to-day basis to spark a vivid conversation. Another helpful aspect was that data owners from the university hospitals were able to provide secure access to a research environment that contained an OMOP database and some of the OHDSI tools, so we were able to directly interact with those components and address relevant aspects such as the data quality of existing data.
Two of our OHDSI experts, Maxim Moinat and Stefan Payralbe, gave the audience a concise overview of all relevant OMOP/OHDSI topics, thus ensuring that attendees from different organisations share a common knowledge base and can further intensify their collaboration successfully in the future.
One of the key sessions of the workshop focused on resources for learning and training and how to best join the growing OHDSI community. Our OMOP experts provided a step-by-step guide to actively participate and learn about the general set up as well as specific components of the OMOP CDM and the OHDSI large scale analytics tool suite.