Considerations of the observational research database utilizing Real World Data

Abstract

Data health initiatives require that we are able to derive evidence from an observational database using Real World Data (RWD) collected exclusively. As the issues related to database design, RWD requires integration into the standard information model and controlled terminologies, inter-organization calibration on such as measurement results, identification and tracking of patient individuals, and bias problems related to data sources. Regarding these issues, standardization of ETL method from RWD, definition of standardized database structure, developing proper methodology to define cohorts, identification of the sources of RWD data and development of quality measures, and ensuring coverage on all cohorts and time series are the highest priority research topics of the medical big data era. We focused on methodologies for defining a common data model for an observational research database and proper phenotyping for the medical informatics concerned. We have investigated the PheKB repository that publishes phenotyping methods in the eMERGE project, and the OMOP Common Data Model and standard vocabulary developed by in OHDSI project. We also propose that rather than trying to implement international CDM in a single step, we should set up two projects. One is promoting the development of a CDM consortium so that electronic medical vendors properly integrate their EHR system and DWH regarding Japanese standard terminologies. The other is that the researcher should develop the mapping tools that map between domestic and international CDM and Terminologies leveraging the CDM.