/gossis

Extracting consistent concepts from multiple databases

Primary LanguageJupyter Notebook

GOSSIS: The Global Open Source Severity of Illness Score

For more information about the GOSSIS consortium, please visit https://gossis.mit.edu/

This repository is focused on the extraction part of the GOSSIS pipeline. It extract consistent concepts from multiple databases. Currently only the ANZICS and eICU-CRD datasets are used to build the GOSSIS-1 model.

For code used to compute GOSSIS predictions, please see: https://github.com/jraffa/rGOSSIS1

Adding variables

  1. Add the info to etc/variable-definitions.yaml
  2. Run python3 variablelist2description.py
  3. Run python3 variablelist2header.py
  4. Add the variable to each load-data.ipynb - sometimes this involves adding it to underlying SQL scripts as well
  5. Re-run everything - if not available, the notebooks should prompt you about it