National Machine Learning Service
Exploreabout National Machine Learning Service
Electronic medical record (EMR) data warehouses for audit, research and surveillance exist in Australia. However, they have variable standards. As the need to utilise health data across organisational boundaries grows, we face challenges regarding data standards, linkage, and governance. Converting EMRs to one common data model (CDM) will open up health data for collaborative use.
To enhance data accessibility for rapid interrogation and evidence generation, this project aims to establish a national, research-ready hospital EMR data asset.
We’ll transform data from hospital EMRs to an international gold standard Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). Tools, mappings and experience gained will be made openly available, and a community of practice and a roadmap for continued national implementation will be built.
By streamlining data governance, consent, and ethics, the data asset will significantly improve the quality, accessibility and feasibility of EMR data warehouses. We’ll convert 3 existing data warehouses from the following centres so that they can be accessed through a single interface:
Learnings from these conversions will guide the conversion of other data warehouse platforms, including the EPIC EMR system, which will be investigated in collaboration with the Peter MacCallum Cancer Centre.
This project will benefit:
This project will deliver:
a national health data asset common data model – health information data is currently siloed in individual EMR systems. The OMOP common data model will harmonise this data and allow direct comparisons. A common data layer will allow standard queries to be run on the data and results shared among researchers
shared data assets enabling research – when a researcher wants to run standard analysis on a number of OMOP-compliant datasets, they will contact the data custodian at each centre to obtain permission. Once permission has been given, a standard analysis script can be provided by the researcher and run on each data set. Results are generated and shared for research, but the actual underlying health information data doesn’t leave the host institution
a workforce skilled in the OMOP common data model – the OMOP common data model is supported by high quality training resources. Extract, transform and load (ETL) workshops are provided by OHDSI Asia Pacific. Researchers are provided hands-on training on preparing their health data for the common data model. The EHDEN Academy provides online training and assessment in the common data model via the learning management system Moodle. The annual OHDSI International Symposium provides training and collaboration opportunities for researchers.
Ultimately, the project will:
For more information, please visit the project website.
You can also read more about the Observational Health Data Sciences and Informatics (OHDSI) Australia and the Observational Health Data Sciences and Informatics.
For training, follow the links below: