Synthetic patient data generated by Synthea can now be loaded into i2b2. The Synthea SyntheticMass sample files have been converted to i2b2-ACT format, and scripts to load Synthea data from scratch are available here: https://github.com/i2b2/i2b2-synthea
Synthea Load Process:
- Set up an i2b2 project with the ACT ontology.
- Either download the SyntheticMass 63k sample in i2b2 format from https://github.com/i2b2/i2b2-synthea/blob/main/syntheamass_63K_sample.zip, or follow the instructions below to load any Synthea dataset from scratch.
Loading Synthea data from scratch
- Download SyntheticMass Data, Version 2 (24 May, 2017)
- All data sets (1k, COVID 10k, COVID 100k) have been verified to work EXCEPT the 100k patients in the large SyntheticMass Version 2 download.
- The 100k patients in the large SyntheticMass Version 2 download needs an extra step to delete invalid records before import. In this case, download synthea_cleanup.pl to your disk, and then run "synthea_cleanup <directory-for-synthea-csv-files>" The fixed csv files will be in <directory-for-synthea-csv-files>/fixcsv.2. Set up an i2b2 project with the ACT ontology.
- Download the scripts from https://github.com/i2b2/i2b2-synthea)
create_synthea_table_<your dbServertype>.sql in your project to create the Synthea tables.
- Import the Synthea data you downloaded in step one into the Synthea tables in your project.
- Load the i2b2-to-SNOMED table in this repository into your project. https://www.nlm.nih.gov/healthit/snomedct/us_edition.html
- Click on the "Download SNOMED-CT to ICD-10-CM Mapping Resources" link to download. (You will need a UMLS account.)
- Unzip the file
- Import the TSV file into a table called SNOMED_to_ICD10 in your database.
- In Postgres and Oracle, follow the additional instructions in the comments at the top of
synthea_to_i2b2_<your dbServerType>.sql to clean up the date formatting.
synthea_to_i2b2_<your dbServertype>.sql to convert synthea data into i2b2 tables (this will truncate your existing fact and dimension tables!)
- Replace references to
i2b2metadata.dbo in the script. Use the database and schema where your ACT ontology tables are.