Overview
AI & Digital Health Innovation is partnering with Medical Informatics Corp. (MIC), the Medical School’s Data Office for Clinical and Translational Research (DOCTR), the University of Michigan HITS, and ITS to store Sickbay waveform and physiologic data in a HIPAA-compliant location for broader use by University of Michigan researchers. Sickbay is an FDA-cleared clinical platform that provides hospitals with a vendor-neutral, integrated healthcare patient-monitoring solution.
This dataset comprises time-sequenced patient waveform and physiologic monitoring data from adult and pediatric ICUs across Michigan Medicine. Data is sourced from the following devices:
| Device Name | Device Type |
| GE Physiological Monitors | Cardiac Monitor |
| PrismaFlex PrisMax | CRRT |
| Fresenius 2008T | Dialysis |
| Fresenius | Critline Dialysis |
| Abbott Centrimag | ECMO |
| Maquet Cardiohelp | ECMO |
| Medtronic INVOS | NIRS |
| Spectrum M | Perfusion Monitor |
| Spectrum Quantum WS | Perfusion Monitor |
| Draeger V500 | Ventilator |
| Hamilton C1 | Ventilator |
| Maquet Servo-u | Ventilator |
| Vyaire Avea | Ventilator |
This data is time-aligned across devices on a subject, and can be associated to a subject across stays and transfers. This data exists in a compact binary form but can be made available in multiple python/Excel/etc.-friendly formats (e.g. CSV, Pickle).
Additional Sickbay waveform and physiologic data are added to the repository each month.
Data Details
| Time period | April 2025 – June 2026 |
| # unique patients | 64,597 |
| # encounters | 91,672 |
| # beds | 1,632 |
| # rooms | 1,502 |
| # departments | 92 |
Getting Started
How to get access
Please review the requirements to access AI&DHI data resources HERE and request access with the data solutions team.
The AI&DHI Data Solutions Managers will meet with you to review the list of targeted patients’ MRNs (and a time frame, if applicable), and then they will copy the appropriate files to the HIPAA-compliant storage location you indicated on your ticket (either HIPAA-compliant Turbo or Michigan Medicine OneDrive).
Navigating the data
File name format is: Bed-ID-epochtime-Device.crf
- The AI&DHI Team will provide Sickbay’s Python packages needed to read the CRF binary files. Python must be on version 3.11 for the packages to be compatible (newer versions will not work).
- Download “sickbay-package.zip” and unzip it to the desired directory.
- Run “sickbay-py-install.bat” by opening the file directly (you may need admin privileges to complete the installation).
- Once complete, import the needed packages and use the CRF module to get started with the files (example of reading a file shown below).
from sickbay import crf
data = crf.CRFReader('path\\to\\file.crf')
print(data)
