Team Ai
Datasetpublic

JimHue/EHRSQL_PostgreSQL_data

EHRSQL PostgreSQL Artifacts This dataset repository hosts database artifacts generated for the companion GitHub repository: https://github.com/Jensen-JinhuiLiu/EHRSQL_PostgreSQL Files eicu.sqlite: original benchmark SQLite database artifact used for eICU. mimic_iii.sqlite: original benchmark SQLite database artifact used for MIMIC-III. postgres_import/eicu/csv/: CSV exports for loading the eICU PostgreSQL database. postgres_import/mimic_iii/csv/: CSV exports for… See the full description on the dataset page: https://huggingface.co/datasets/JimHue/EHRSQL_PostgreSQL_data.

sourceHugging Facecc-by-4.0updated 2mo agoView on Hugging Face
0likes80downloads
Dataset Card

EHRSQL PostgreSQL Artifacts

This dataset repository hosts database artifacts generated for the companion GitHub repository:

https://github.com/Jensen-JinhuiLiu/EHRSQL_PostgreSQL

Files

  • —eicu.sqlite: original benchmark SQLite database artifact used for eICU.
  • —mimic_iii.sqlite: original benchmark SQLite database artifact used for MIMIC-III.
  • —postgres_import/eicu/csv/: CSV exports for loading the eICU PostgreSQL database.
  • —postgres_import/mimic_iii/csv/: CSV exports for loading the MIMIC-III PostgreSQL database.

The companion GitHub repository contains the preprocessing scripts, PostgreSQL schema/load files, generated benchmark JSON files, and a checksum manifest for these artifacts.

Download

Download directly into the companion GitHub repository root:

bash
hf download JimHue/EHRSQL_PostgreSQL_data --repo-type dataset --local-dir .

Or clone this dataset repository separately:

bash
git clone https://huggingface.co/datasets/JimHue/EHRSQL_PostgreSQL_data

Loading PostgreSQL

After downloading the artifacts into the GitHub repository root, follow:

  • —postgres_import/README.md
  • —postgres_import/create_databases.sql
  • —postgres_import/eicu/schema_no_fk.sql
  • —postgres_import/eicu/load_data.sql
  • —postgres_import/mimic_iii/schema_no_fk.sql
  • —postgres_import/mimic_iii/load_data.sql

Attribution

These artifacts are derived from the EHRSQL benchmark:

EHRSQL: A Practical Text-to-SQL Benchmark for Electronic Health Records https://github.com/glee4810/EHRSQL