Every month, we offer several of our data partners the opportunity to present themselves, how far they are in their mapping process, the lessons they learned during their mapping and which research studies they are a part of.
Asst Prof Selçuk Şen, Istanbul University, Istanbul Faculty of Medicine, Dept of Medical Pharmacology, explores their experience in mapping COVID-19 clinical data of 59,626 patients from their Oracle system to the OMOP CDM with the technical assistance of the EHDEN COVID-19 ETL Taskforce.
Prof Şen outlines the process from initial workshops and utilising tools such as Rabbit in a Hat, Usagi, the Data Quality Dashboard, and Achilles.
Following the end of the mapping cycle, an inspection report, a series of surveys into the process and an entry into the developing EHDEN Data Partner Catalogue were completed. Finally, next steps and proposals to be involved in relevant network studies are outlined.
Filip Malković of Heliant, presents the University Clinical Centre (UCC) of Serbia’s experience in mapping COVID-19 clinical data, in conjunction with Clinerion, an EHDEN certified SME.
He covers the background of the UCC, ETL process in conjunction with the EHDEN COVID-19 ETL Taskforce, the vocabulary mappings, data quality assessment, cohorts and studies performed and next steps planned.
Filip explores the use of the relevant tools, Rabbit in a Hat, Usagi, Data Quality Dashboard, and Achilles, used in the mapping of COVID-19 patients within an 800,000 patient records dataset from 2019 onwards, and some of the studies performed on top of the CDM-mapped dataset, such as an initial feasibility study, and an Adverse Events of Special Interest (AESI) R package run for population trends with regards to COVID-19 vaccines.
Dries Hens and Wout Olyslagers of Lynxcare, an AI-powered clinical data platform, and working to be an EHDEN certified SME, explored the mapping of COVID-19 records data across 3 different EHR systems in 6 different Belgium hospital data sources.
The emphasis was the deployment of a clinical decision support platform on top of the federated CDM-mapped hospital EHRs, inclusive of using natural language processing models to incorporate unstructured data. Wout discussed the ETL cycle, and particular challenges and solutions, inclusive of file formats to data interpolation, and the incorporation of NLP outputs into the OMOP CDM.
Lynxcare now have an automated refresh process, inclusive of the ETL and analytical OHDSI tools in docker environments with a preconfigured SQL query library. Dries, Wout and colleagues are looking forward to being EHDEN certified, but also collaborating in future research.