Decision curve analysis is a statistical method that evaluates modelsand tests in terms of their clinical consequences. This is unliketraditional accuracy measures - such as the area-under-the-curve orBrier score - which do not take into account considerations such as, forinstance, it being worse to miss a cancer (false negative) than do anunnecessary biopsy (false positive). Decision curve analysis evaluatesthe net benefit of a model or test in comparison to the two defaultstrategies of treat all patients and treat no patients.

In the “Software Tutorial” tab, we provide step-by-step instructionsfor code installation and then actually performing decision curveanalysis for binary and time-to-event outcomes. Separate instructionsare given for R, Stata, SAS and Python. We also cover multivariabledecision curve analysis, evaluation of published models, saving netbenefit values, presenting results in terms of interventions avoided,survival outcomes, competing risks, case-control designs andincorporating harms.

In the “Peer-Reviewed Literature” tab, you can find references to theoriginal papers describing decision curve analysis methodology, theorypapers explaining the mathematical derivation of net benefit,introductory guides (including separate guides for researchers andreaders) and examples of editorials that recommend the use of decisioncurve analysis.

If you can’t find an answer to a question in one of the other tabs,click on the “Discussions” tab, which will navigate you to the GitHubdiscussions forum for all Q&A. If you post a question, someone fromour team will get back to you.