MIMIC-IV: Clinical Cohort Analysis#

A series of tutorials walking through a complete clinical analysis pipeline on the MIMIC-IV demo dataset (100 de-identified patients from Beth Israel Deaconess Medical Center).

Each script is self-contained and downloads the database automatically on first use via access().

The demo database covers:

  • Patient demographics (gender, age)

  • Hospital admissions with types and locations

  • Medical procedures coded with ICD standards

  • Pharmacy prescriptions and medication data

Data access

Note

SQL ingestion requires connectorx: pip install 'tanat[sql]'

Exploring a patient cohort

Exploring a patient cohort

Filtering and preparing a cohort

Filtering and preparing a cohort

Analysing and clustering cohort sequences

Analysing and clustering cohort sequences

Survival analysis from cohort clusters

Survival analysis from cohort clusters

Learning clinical temporal patterns with SWoTTeD

Learning clinical temporal patterns with SWoTTeD