We work with you to bring your health data from EHR systems, registries, observational data records, and other health data sources onto a standard platform or common data model, such as OMOP CDM.
Data harmonization involves transferring data from a source system, often a proprietary one, to a common data representation, such as OHDSI’s OMOP Common Data Model (CDM). This process can vary in complexity depending on how the source data is structured, how the information is coded (or not coded), language, volume of data, and other factors. It is therefore essential that a structured approach is used, with the necessary commitment by all participants.
There are several steps needed to get the source data transformed to the target data model.
- The first step is to analyze the source data, in order to understand how the source data is organized, what coding systems are used, and what data is available. The mapping logic that will then be defined between the source data and the target platform (OMOP CDM or other) is based on this source data analysis, so it is crucial that the data owner and representatives (clinical, technical, and others) are involved and fully committed to this process.
- Depending on what coding systems are used, mapping the codes used in the data source to those used in the target may be straight forward (e.g. ICD10CM → SNOMED), or a more complex and longer process (e.g. in-house codes/descriptions in local language → LOINC).
- When mapping logic and code mappings are available, the actual ETL (Extract – Transform – Load) script can be designed, implemented and tested – this is the code that will load the data from the source to the target, applying the mapping and logic defined.
- The ETL can then be executed by the data owner in their environment, and the quality of the process assessed – any adjustments and tweaks needed can be done, and an updated ETL again tested and assessed. Once the data owner approves the ETL and quality assessment, the ETL can be deployed and executed on the actual data at the data owner’s site.
- Depending on the assessment of available infrastructure and resources, dedicated infrastructure (e.g. servers) may need to be set up to facilitate testing and deployment.
- Finally, in order to get the most out of the harmonized data and related tools, training may be needed depending on familiarity with the target platform, such as OMOP CDM and OHDSI tools.
What can we do?
edenceHealth can perform all of the above tasks in order to implement your data harmonization project. We can also perform selected steps in collaboration with your in-house experts, perform reviews of planned or already executed tasks, and do data quality assessments both before and after a data harmonization process.
HONEUR data harmonization
We are also doing OMOP CDM data harmonization projects for the Haematology Outcomes Network in Europe (HONEUR) initiative, a Janssen-led collaboration to create a federated network of Haematology centers across Europe and beyond.
Other data harmonization projects
Our team has experience from multiple data harmonization projects towards the OMOP CDM platform, as well as extensive experience working with life science data, rare diseases data, medical claims data, and ETL projects in general.