The purpose of the Consortium for Oral Health Related Informatics is to establish an Oral Health Data Repository that can accept and integrate data from disparate dental data sources, and allow end users to explore and extract information to support their specific research or decision making needs.
We are working with UT, UCSF, Harvard, and Tufts to create a joint data warehouse that contains dental data from each institution. In support of a G08 grant through Dr. Muhammad Walji, the BIG team is working to create a set of scripts which will extract data from Axium instances (running at affiliate sites) and convert this to a common format. The scripts will be distributed to the participating institutions where member's will execute the scripts and send the extracted data to UT for the BIG team to load into a combined COHRI i2b2 instance, allowing for the querying of data across institutions.
Subcomponents of the project will include a repeatable process for data extraction, transmission and loading and a process for authenticating and authorizing user access to the COHRI i2b2 instance. Both processes will be documented and distributed to the Consortia members.
Finally, a front end website will be provided to COHRI members that will allow faculty at participating institutions to login using Shibboleth to support federated authentication once granted access from Dr. Walji. It will allow users to use their local institutional userid and passwords to authenticate to the application and query the data repository.
UT Health has created a set of scripts to extract data from the Axium database. These scripts create csv files which correspond to the i2b2 table, observation fact table, and patient dimension tables. We have distributed these scripts to our four partner institutions to run against their production Axium databases and extract the demographics data, Medical and Dental History forms, Procedures, Providers, Diagnosis, Odontogram and Perio charting data. Once the extraction is complete, the institutions are sending their data to UT Health and we are importing the data into our i2b2 COHRI data warehouse. In addition, if the institution already has an i2b2 infrastructure, these files can be loaded into their local i2b2 data warehouses to give them access to their own data through i2b2.
We will be working to map the various terminologies from the participating sites to what will become the COHRI terminology. This will be both a manual and semi-automated process to map the terms together. In addition to the mapping task, we will be working to add additional data elements to the Axium to i2b2 data extract scripts. Eventually, we plan to expand the group of participating institutions to the entire COHRI Consortium.