Extensions to decision curve analysis, a novel method for evaluating diagnostic tests, prediction models and molecular markers
2008

Extensions to Decision Curve Analysis for Evaluating Diagnostic Tests

Sample size: 740 publication Evidence: moderate

Author Information

Author(s): Andrew J Vickers, Angel M Cronin, Elena B Elkin, Mithat Gonen

Primary Institution: Memorial Sloan-Kettering Cancer Center

Hypothesis

Can decision curve analysis be effectively extended to improve the evaluation of diagnostic tests, prediction models, and molecular markers?

Conclusion

Decision curve analysis can be easily extended to many applications common to performance measures for prediction models.

Supporting Evidence

  • Decision curve analysis combines accuracy measures with clinical applicability.
  • Simulation studies showed that repeated 10-fold cross-validation provided the best method for correcting a decision curve for overfit.
  • Decision curve analysis allows assessment of clinical relevance without needing additional data.

Takeaway

This study shows a new way to evaluate medical tests that helps doctors make better decisions without needing extra data.

Methodology

The study presents extensions to decision curve analysis including correction for overfit, confidence intervals, application to censored data, and calculation of decision curves directly from predicted probabilities.

Potential Biases

Models evaluated on the same data set used to build them are at risk for overfitting.

Limitations

The study does not provide methods for correcting decision curve analysis for overfit or for calculating confidence intervals.

Participant Demographics

The study involved 740 men undergoing biopsy for prostate cancer.

Digital Object Identifier (DOI)

10.1186/1472-6947-8-53

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication