Janos Hajagos, PhD; Chief of Data Analytics, Research Assistant Professor, Department of Biomedical Informatics, Stony Brook University
Wednesday, Mar 27, 2019 3pm—4pm BMI Conference Room HSC-L3 Room 045
Data extracted from EHRs (Electronic Health Records) and administrative billing/claims systems are stored in different relational database schemas. Common Data Models (CDMs) are designed to align data from different sources into a single database schema for analysis. The OHDSI (Observational Health Data Sciences and Informatics) CDM is increasingly being used in informatics. I will introduce the OHDSI CDM and discuss how it maps coded data into source and mapped concepts. As an example, I will show the mapping of two different local sources into the OHDSI CDM. Once data is transformed into the CDM relational data model, it can be automatically transformed for building and testing machine learning models.