Sepsis Survival Clinical Records from Norway and South Korea

110,204 patient admissions from a primary cohort in Norway, plus a 19,051-admission study cohort and a 137-patient validation cohort from South Korea. The dataset, created by Chicco and Jurman (2020), contains minimal clinical records for patients with sepsis-related conditions, with all categorical variables pre-encoded.

Use Cases

• Predicting sepsis survival likelihood based on minimal clinical features.
• Validating sepsis prediction models on an external cohort from a different geography.
• Studying the generalizability of clinical models across patient populations.
• Analyzing the relationship between patient admissions and sepsis outcomes.

Strengths

• Contains a large primary cohort of 110,204 patient admissions.
• Includes an external validation cohort of 137 patients from a different region (South Korea) for testing generalizability.
• Data has been preprocessed with categorical variables encoded, reducing initial preparation work.
• Dataset explicitly has no missing values.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.
• The dataset contains sensitive information about patient gender and age, requiring careful handling.

Provenance

Source

OpenML, with data from a primary and study cohort in Norway and a validation cohort from South Korea.

Collection Method

Likely extracted from clinical records for a research study on sepsis survival.

Time Range

null

Freshness

Last updated date is unknown.

Geography

Norway (primary and study cohorts) and South Korea (validation cohort).