MIMIC-IV, a freely accessible electronic health record dataset

Area of Science:

• Health Informatics
• Clinical Research Data Management
• Biomedical Data Science

Background:

• Digital health data collection is common in hospitals, offering research potential.
• Existing clinical data archival systems are often inaccessible and not optimized for research analysis.
• Barriers exist in accessing and utilizing clinical data for research purposes.

Purpose of the Study:

• To present MIMIC-IV, a publicly available database of deidentified electronic health records.
• To provide a resource that supports diverse clinical research studies and educational initiatives.
• To reduce obstacles hindering clinical research by offering accessible, structured data.

Main Methods:

• The MIMIC-IV database was compiled from electronic health records at Beth Israel Deaconess Medical Center.
• Data includes deidentified patient measurements, orders, diagnoses, procedures, treatments, and clinical notes.
• The database is structured to enhance interpretability and analysis for research applications.

Main Results:

• MIMIC-IV offers a comprehensive dataset from routine clinical practice.
• The database includes a wide range of clinical information, including free-text notes.
• It is designed to be publicly accessible to researchers.

Conclusions:

• MIMIC-IV significantly lowers barriers to conducting clinical research.
• The database serves as a valuable resource for advancing medical research and education.
• Accessible electronic health record data is crucial for future healthcare innovations.