> For the complete documentation index, see [llms.txt](https://docs.bcbi.brown.edu/codiac-for-health/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.bcbi.brown.edu/codiac-for-health/ursa-ri/datasets/mimic-iv.md). # MIMIC-IV ## Overview The Medical Information Mart for Intensive Care (MIMIC) is a publicly available repository of deidentified electronic health record (EHR) data from patients admitted to the Beth Israel Deaconess Medical Center (BIDMC) in Boston, Massachusetts. This resource was created in collaboration with Massachusetts Institute of Technology to lower barriers to reproducible clinical research by making real world hospital data accessible in a deidentified form. MIMIC is organized into four datasets, which altogether contain five modules. The following info pertain to their latest versions: MIMIC-IV (version 3.1): This dataset contains info derived from 2008-2022. * The Hosp module contains hospital-level tables related to 546,028 hospitalizations for 223,452 unique individuals. * The ICU module is for for intensive care unit admissions and contains 94,458 ICU stays for 65,366 unique individuals MIMIC-IV-ED (version 2.2): * The ED module contains tables related to emergency department (ED) visits from 2011 to 2019. There are 425,087 ED visits of which 203,016 led to hospital admissions. MIMIC-IV-Note (version 2.2): * The Note module contains tables related to 331,794 deidentified discharge summaries from 145,915 patients admitted to the hospital and emergency department. MIMIC-CXR (latest version 2.0) : * The CXR module contains tables related to 377,110 images corresponding to 227,835 radiographic studies in Digital Imaging and Communications in Medicine (DICOM) format with free text radiology reports made in the emergency department from 2011 to 2016 Other notable modules include the following: #### MIMIC-IV-ECG * The ECG module contains approximately 800,000 diagnostic electrocardiograms from nearly 160,000 unique patients. These diagnostic ECGs use 12 leads, are 10 seconds in length and are sampled at 500 Hz. #### MIMIC-IV-ECHO * The ECHO module contains 206,488 echocardiogram studies (including 179,928 transthoracic, 16,389 stress, and 10,171 transesophageal echocardiograms) from 91,372 unique patients between 2008 and 2022. * The MIMIC-IV-ECHO module contains structured echocardiographic measurements and DICOM files from echocardiography exams. ## Methodology and Generation ### **Acquisition** Data across the MIMIC-IV family were sourced from Beth Israel Deaconess Medical Center's (BIDMC) systems, with each component drawn from their respective sources. For instance, MIMIC-IV Core drew from the hospital-wide EHR and ICU bedside systems, excluding patients under 18 at first visit or on an enhanced-protection list; MIMIC-IV-Note included only notes within one year of a hospital or ED encounter; MIMIC-IV-ED drew emergency department records in XML format; and MIMIC-CXR separately sourced chest radiograph studies via the local EHR and Radiology Information System (RIS), alongside their associated free-text reports. ### **Transformation** Raw data were converted into structured, analysis-ready formats while preserving the integrity of the original clinical record. MIMIC-IV was transformed via custom SQL into a schema organized into the Hosp and ICU modules, designed for backward compatibility with MIMIC-III, with tables exported as CSV files and no data cleaning applied so the dataset reflects real-world clinical data as recorded. Out-of-hospital deaths were linked to Massachusetts state vital records using a custom workflow combining exact and fuzzy matching (name, date of birth, social security number), prioritizing sensitivity over specificity. MIMIC-IV-ED's XML extracts were converted into a denormalized relational database to support analysis. MIMIC-CXR integrated three separately processed streams into a single unified dataset: EHR data, DICOM-format imaging, and extracted radiology reports (stripped of administrative/clinical metadata). ### **Deidentification** All datasets meet HIPAA Safe Harbor requirements, though each applies methods suited to its data type. Structured data across MIMIC-IV, MIMIC-IV-ED, and MIMIC-CXR had direct identifiers removed and replaced with randomly assigned IDs (subject\_id, hadm\_id, stay\_id) that remain consistent across datasets to enable cross-linkage. Dates were shifted using patient-specific offsets to preserve within-patient chronological relationships. Free-text and unstructured data required additional handling: MIMIC-IV-Note and MIMIC-CXR's radiology reports were deidentified using a combination of rule-based methods, a purpose-built neural network, and manual review; MIMIC-IV-ED's free-text fields were scrubbed and replaced with placeholders; and MIMIC-CXR's DICOM imaging had PHI removed from both metadata and pixel data via a custom algorithm with manual validation. Finally, released datasets were exported in a character-based, comma-delimited format for distribution. ### Derived tables The MIMIC team provides and maintains code in the public mimic-code repository () to generate tables and common variables such as severity scores to make analysis easier. ## Data Access Access to MIMIC is provided via Physionet and restricted. For Brown University staff and students, access can be provided on OSCAR after meeting the following requirements. Researchers generally need to be credentialed, complete human subjects training via CITI courses and sign a data use agreement before obtaining the full dataset. For more detailed instructions, please refer to your Primary Investigator or Brown faculty. ## Standardized Healthcare Vocabularies ### Diagnoses and Procedures * ICD-9-CM diagnoses * ICD-9-PCS procedures * ICD-10-CM diagnoses * ICD-10-PCS procedures * HCPCS ### Medications * National Drug Code (NDC) ### Labs * Logical Observation Identifiers, Names and Codes(LOINC) ### Imaging * DICOM ## References 1. [Health Insurance Portability and Accountability Act \[HIPAA\] of 1996](https://www.govinfo.gov/app/details/PLAW-104publ191), Pub. L. No. 104-191. 2. [CDC ICD-10-CM Browser Tool](https://icd10cmtool.cdc.gov/?fy=FY2026) 3. [WHO ICD-10 CM Site ](https://icd.who.int/browse10/2019/en) 4. [ICD 9-CM Look up Site ](https://www.icd9data.com/) ## Resources ### Articles * Johnson AEW, Bulgarelli L, Shen L, Gayles A, Shammout A, Horng S, Pollard TJ, Hao S, Moody B, Gow B, Lehman LH, Celi LA, Mark RG. [MIMIC-IV, a freely accessible electronic health record dataset.](https://pubmed.ncbi.nlm.nih.gov/36596836/) Sci Data. 2023 Jan 3;10(1):1. doi: 10.1038/s41597-022-01899-x. Erratum in: Sci Data. 2023 Jan 16;10(1):31. doi: 10.1038/s41597-023-01945-2. Erratum in: Sci Data. 2023 Apr 18;10(1):219. doi: 10.1038/s41597-023-02136-9. PMID: 36596836; PMCID: PMC9810617. * Johnson AEW, Pollard TJ, Berkowitz SJ, Greenbaum NR, Lungren MP, Deng CY, Mark RG, Horng S. [MIMIC-CXR, a de-identified publicly available database of chest radiographs with free-text reports.](https://pubmed.ncbi.nlm.nih.gov/31831740/) Sci Data. 2019 Dec 12;6(1):317. doi: 10.1038/s41597-019-0322-0. PMID: 31831740; PMCID: PMC6908718. * Goldberger AL, Amaral LA, Glass L, Hausdorff JM, Ivanov PC, Mark RG, Mietus JE, Moody GB, Peng CK, Stanley HE. [PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals.](https://pubmed.ncbi.nlm.nih.gov/10851218/) Circulation. 2000 Jun 13;101(23):E215-20. doi: 10.1161/01.cir.101.23.e215. PMID: 10851218. ### Links * [MIMIC-IV v3.1 Physionet Page](https://physionet.org/content/mimiciv/3.1/) * [MIMIC-IV Physionet Search Page for all module related info](https://physionet.org/content/?topic=MIMIC-IV) * [MIMIC MIT website](https://mimic.mit.edu/docs/IV/) * [MIMIC Code Repository](https://github.com/MIT-LCP/mimic-code/) * [MIMIC-CXR Physionet Page](https://physionet.org/content/mimic-cxr/2.1.0/) * [MIMIC-IV-ED Physionet Page](https://physionet.org/content/mimic-iv-ed/2.2/) * [MIMIC-IV-Note Physionet Page](https://physionet.org/content/mimic-iv-note/2.2/) * [MIMIC-IV-ECG Physionet Page](https://physionet.org/content/mimic-iv-ecg/1.0/) * [MIMIC-IV-ECHO Physionet Page](https://physionet.org/content/mimic-iv-echo/1.0/) * [MIMIC-IV Datasets in Oscar](https://docs.google.com/document/d/1ZkQf-DaBp3zDBOK383OoAjLSolArP71IG4YKhK0GbdQ/edit?tab=t.0) * [DICOM standard](https://www.dicomstandard.org/current) * [Studies that have utilized MIMIC-IV on PubMed](https://pubmed.ncbi.nlm.nih.gov/?term=mimic-iv)

--- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.bcbi.brown.edu/codiac-for-health/ursa-ri/datasets/mimic-iv.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.