Elamur Research
☰
home / papers / 2023 / MIMIC-IV, a freely accessible electronic health record dataset

MIMIC-IV, a freely accessible electronic health record dataset

Written for the Research library (claude-opus-5-5) from the paper text. Original — nature.com/articles/s41597-022-01899-x.pdf

Summary

Hospitals record large amounts of clinical data digitally, but it usually sits in archival systems built for storage, not research, and is hard for outside scientists to reach or interpret. Existing public critical care datasets also tend to be dated, cover only one data type, or need regular refreshing. The authors, a team from MIT and Beth Israel Deaconess Medical Center (BIDMC), release MIMIC-IV to address this. It contains deidentified records of adults treated in the emergency department or an ICU at BIDMC between 2008 and 2019. Data from the hospital warehouse and the MetaVision ICU system were moved to PostgreSQL, restructured with SQL scripts, and stripped of protected health information. The data are organized into linkable modules (hosp, icu, note) and enriched with external coding systems and state death records. The result is a modern, well-validated resource that supports research, teaching, and tool development. It also connects to earlier MIMIC releases and can be extended to new data types.

Key points

  • Scale (v2.2): 431,231 hospital admissions from 180,733 patients, plus 73,181 ICU stays from 50,920 patients. In-hospital mortality is 2.1% for all admissions and 11.6% for ICU stays.
  • Structure: The relational modules hosp, icu, and note are linked through subject_id, hadm_id, and shifted timestamps. New tables include emar, poe, omr, and pharmacy. The icustay_id field was replaced by stay_id.
  • Mortality after discharge: Instead of the SSDMF, the RecordLinkage library was used to link patient records with death data from Massachusetts' official vital records registry. This linked 65,805 of 150,832 candidate individuals.
  • Quality checks: Unit tests covered integrity, consistency, and deidentification. For example, at least 99% of ICU stays had to carry a billed diagnosis and a heart rate measurement.
  • Access: Users must complete human-subjects training and sign a DUA. The MIMIC Code Repository supplies loaders and derived tables. The MIMIC-III CareVue subset covers the years 2002–2008.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Other

Cited by

114 papers of our library cite it (of the 140673 whose bibliographies we read).

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.1038/s41597-022-01899-x

@inproceedings{johnson2023mimic,
  title = {{MIMIC-IV, a freely accessible electronic health record dataset}},
  author = {Johnson, Alistair E. W. and Bulgarelli, Lucas and Shen, Lu and Gayles, Alvin and Shammout, Ayad and Horng, Steven and Pollard, Tom and Hao, Sicheng and Moody, Benjamin and Gow, Brian J. and Lehman, Li-wei H. and Celi, Leo Anthony and Mark, Roger G.},
  booktitle = {Scientific Data},
  year = {2023},
  doi = {10.1038/s41597-022-01899-x},
  url = {https://doi.org/10.1038/s41597-022-01899-x}
}

References

10 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Moody, G. B. & Mark, R. G. A database to support development and evaluation of intelligent intensive care monitoring. In Computers in Cardiology 1996, 657–660 (IEEE, 1996). find
  2. Saeed, M. et al. Multiparameter intelligent monitoring in intensive care ii (MIMIC-II): a public-access intensive care unit database. Critical care medicine 39, 952 (2011). find
  3. Johnson, A. E. et al. MIMIC-III, a freely accessible critical care database. Scientific data 3, 1–9 (2016). find our card
  4. Pollard, T. J. et al. The eICU collaborative research database, a freely available multi-center database for critical care research. Scientific data 5, 1–13 (2018). find
  5. Thoral, P. J. et al. Sharing ICU patient data responsibly under the society of critical care medicine/european society of intensive care medicine joint data science collaboration: the amsterdam university medical centers database (AmsterdamUMCdb) example. Critical care medicine 49, e563 (2021). find
  6. Hyland, S. L. et al. Early prediction of circulatory failure in the intensive care unit using machine learning. Nature medicine 26, 364–373 (2020). find
  7. Faltys, M. et al. Hirid, a high time-resolution ICU dataset (version 1.1.1). PhysioNet (2021). find
  8. Zeng, X. et al. PIC, a paediatric-specific intensive care database. Scientific data 7, 1–8 (2020). find
  9. Fontaine, D. PGLoader. ver. 3.6.2. https://github.com/dimitri/pgloader (2020). ↗
  10. MassHealth. Special notices for acute hospitals. https://www.mass.gov/service-details/special-notices-for-acute-hospitals (2020). 11. for Medicare & Medicaid Services, C. ICD-9-CM Diagnosis and Procedure Codes: Abbreviated and Full Code Titles. https://www. cms.gov/Medicare/Coding/ICD9ProviderDiagno… — [doi:10.5281/zenodo.6470616](https://doi.org/10.5281/zenodo.6470616) ↗