Thank you! @schuemie
Merck and Amazon are working together to develop a tool that can translate user’s question to SQL query and directly retrieve the result from OMOP data.
It has a similar concept with Criteria2Query Criteria2Query, but it is a deep learning model with higher accuracy and flexibility. The workflow does not depend on Atlas.
We also have a set of training data (user-generated questions and corresponding SQL queries) we can include in the github repository. The training data can be used for more general Text-to-SQL purpose.
Can we call it “NL2SQL-OMOP”? (Changing it to NOSTOS, “navigate omop-structured data via text-to-sql" based on suggestion from Chris)
The github users are: miguelrc1, liyupeng111, Weilin37.