MIMIC-III, short for The Medical Information Mark for Intensive Care III, is a large, publicly available dataset of de-identified medical records, including vital signs, medications, laboratory measurements, patient signs recorded by nursing staff, fluid balance, procedure codes, diagnostic codes, imaging reports, length of stay, survival data, etc.

This dataset can be used in academic and industrial research, service quality improvement, and higher education courses.

The dataset includes:

• 112,000 clinical reports (average length 709.3 tokens)
• 1,159 top ICD-9 codes
• Each report was assigned an average of 7.6 codes.

Example image: