MIMIC-IV-on-FHIR
summary
Converts the 300,000‑patient MIMIC‑IV Postgres EHR into FHIR using PostgreSQL, Python and HAPI FHIR, outputs validated NDJSON bundles and publishes a demo 100‑patient set on PhysioNet; workflow uses a pi-mimic-fire package for extraction, validation and export and a SUSHI/FHIR Shorthand IG for profiles and 20+ code systems. Handles transfers by storing care units in encounter.location, represents microbiology as a tiered observation with linked susceptibilities, and models complex meds with a single medication resource containing multiple ingredients; several microbiology/lab extensions are added.
AI-written summary · about
chapters (30)
- 0:00Introduction
- 0:38Collaborators
- 0:52Learning Objectives
- 1:14What is MIMIC
- 2:07What MIMIC looks like
- 3:44Motivation to convert
- 4:25Goals for converting
- 5:01Workflow
- 5:33Implementation Guide
- 6:04Generating Terminology
- 6:55Binding Terminology to Profiles
- 7:46Profiling
- 8:25Profile View
- 9:13Implementing Guide
- 9:24Mapping Mimic to FHIR
- 9:58Mapping Process
- 10:32Mapping Examples
- 11:44Care Units as Locations
- 12:32Microbiology Mapping
- 13:25Observation Microtest
- 14:08Medication Mapping
- 15:05Medication Resource
- 15:44Medication Request
- 16:38Global View
- 16:58Validation
- 18:13Pi MimicFire
- 19:03Published
- 19:37Data
- 20:55Download Data
- 28:39Summary
