MIMIC-IV (Medizinisch Information Mart for Intensive Care IV) is a comprehensive and publicly accessible database that contains de-identified health data associated with critical care patients. It is designed to support research in various fields, including healthcare analytics, maschinellem Lernen, and clinical decision-making.
The MIMIC-IV database includes a wide array of information collected from patients admitted to intensive care units (ICUs). This includes vital signs, laboratory test results, medications, diagnostic codes, and more. The dataset is sourced from the Beth Israel Deaconess Medical Center (BIDMC) in Boston, Massachusetts, and spans multiple years, providing a rich longitudinal perspective on patient care and outcomes.
One of the key features of MIMIC-IV is its structured format, which allows researchers to easily access and analyze the data. The database is organized into tables that contain demographic information, clinical notes, and time-series data on patient vitals and interventions. This structured approach facilitates various types of analyses, including prädiktiven Modellierungen, cohort studies, and epidemiological research.
MIMIC-IV baut auf seinem Vorgänger, MIMIC-III, auf und enthält Verbesserungen in Datenqualität, volume, and the inclusion of additional patient demographics. It serves as a vital resource for clinicians, researchers, and data scientists who aim to improve patient outcomes, understand disease patterns, and develop innovative healthcare solutions.
To access MIMIC-IV, users must complete a training course on human subjects research and sign a data use Vereinbarung, die die ethische Nutzung sensibler Gesundheitsinformationen sicherstellt.