dcurves: Decision Curve Analysis for Clinical Model Evaluation

Daniel D. Sjoberg's dcurves dataset provides data for evaluating and comparing diagnostic and prognostic models using decision curve analysis. This method incorporates clinical consequences and can be applied to models with continuous or dichotomous results, requiring only the test dataset. The approach is based on established methods from Vickers (2006, 2008) and Pfeiffer (2020).

Use Cases

• Evaluating the clinical net benefit of a prediction model based on the described decision curve analysis method.
• Comparing multiple prognostic models on a single test dataset to incorporate clinical consequences.
• Assessing models with continuous outputs using a technique that does not require additional data collection.
• Applying decision-analytic techniques for model validation as referenced in the provided citations.

Strengths

• Dataset is specifically designed for the established decision curve analysis method, which is referenced by three peer-reviewed publications.
• The method can evaluate both continuous and dichotomous model outputs using only the test dataset.

Limitations

• Row count, column definitions, and sample data are unavailable, limiting suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.
• Last update date and license are unknown.

Provenance

Source

Daniel D. Sjoberg via paperswithcode.

Collection Method

Likely contains data for demonstrating decision curve analysis methods.