CogStack Dashboards Demo Scripts

This repository demonstrates figures created by data stored in CogStackDashboards.

This repo is intended to be a guide on how to access the dashboards and includes example scripts that can be used to gain clinical insights from the MIMIC-IV dataset. Along with additional capacity for querying annotations via discharge_annotations and radiology_annotations.

├── data/
│   ├── co_occurrences/ - Co-occurrence information for SNOMED CT concepts
│   └── SNOMED/ - Required SNOMED CT Files, See below
├── figures/ - Images used in the paper and generated from scripts
├── sample_access_scripts/ - Simple demo scripts showing how to request data via curl or Python
├── staging_scripts/ - Staging scripts to save data from CogstackDashboards in local files
├── visualisation_scripts/ - Scripts used to generate images used in the paper
└── README.md

Access:

1. Sign up for a PhysioNet account
2. Complete training and obtain credentialed status
3. Log into CogStackDashboards with your PhysioNet account
4. For API access, see sample_access_scripts for examples of how to create appropriate queries

Prerequisite: SNOMED CT Required Files

SNOMED CT is required for resolving higher-level ancestors of top-level nodes and translating SNOMED codes into human-readable names.

International SNOMED CT Downloads

https://www.nlm.nih.gov/healthit/snomedct/international.html

UK SNOMED CT Downloads

https://isd.digital.nhs.uk/trud/users/guest/filters/0/home

After downloading, extract the following two files from the ontology package. They follow this naming convention:

• sct2_Description_Full-en_INT_YYYYMMDD.txt
• sct2_Relationship_Full_INT_YYYYMMDD.txt

Save them in data/SNOMED/ as:

• sct2_description.txt
• sct2_relationship.txt