AUC Values for Mortality Prediction Models 2014–2023

AUC values from ROC analysis for predicting short-term mortality (<12 hours) and 30-day mortality, covering the decade from 2014 to 2023. The dataset was authored by Ingo Gräff and is available on figshare. It is a small dataset, 5.5 KB in size, stored in an XLS file.

Use Cases

• Benchmark new mortality prediction models against AUC values for short-term mortality from 2014–2023.
• Analyze temporal trends in model performance for 30-day mortality prediction using the decade-long AUC data.
• Compare the discriminative ability (AUC) of different clinical risk scores for short-term vs. 30-day mortality outcomes.

Strengths

• Provides a decade-long time series (2014–2023) of model performance metrics.
• Focuses on two clinically distinct mortality endpoints: short-term (<12 hours) and 30-day.

Limitations

• Extremely small data volume (5.5 KB), indicating a limited number of AUC data points.
• Lacks underlying patient-level data, limiting analysis to aggregate performance metrics.
• Unknown row count prevents assessment of the dataset's comprehensiveness.

Provenance

Source

Ingo Gräff via figshare.

Collection Method

Presumably compiled from ROC analyses of clinical prediction models.

Time Range

2014–2023

Freshness

Last updated in April 2026, indicating recent curation.