SCIENCE@home

ROC curves for clinical prediction models part 1: ROC plots showed no added value above the AUC when evaluating the performance of clinical prediction models.

J. Verbakel,E. Steyerberg,H. Uno,B. De Cock,L. Wynants,G. Collins,B. Van calster

Published 2020 in Journal of Clinical Epidemiology

ABSTRACT

BACKGROUND Receiver operating characteristic (ROC) curves are widely used in reports on clinical risk prediction models to demonstrate the ability of a model to discriminate between patients with and without a certain condition. OBJECTIVE We discuss the limited value of ROC curves in this context, and state that curves without threshold information are uninformative. RESULTS Based on a pragmatic search strategy, 52 out of 86 contemporary medical publications (60%) presented uninformative ROC curves. By not plotting thresholds on the curve, most ROC curves provided limited information over and above the area under the curve (AUC) as a summary of discriminatory ability. If a graphical assessment of discriminative ability is desired, we recommend classification plots over ROC curves. These present sensitivity and specificity conditional on risk threshold, which is key to decision making. Classification plots can be supplemented with measures such as net benefit to assess the potential clinical utility of a model. CONCLUSION We recommend focusing on the AUC rather than on the whole ROC curve; using classification plots if a visualization of discriminatory ability is desired; and sensitivity, specificity and net benefit at a range of clinically relevant risk thresholds for assessment of clinical utility.

PUBLICATION RECORD

CITATION MAP

EXTRACTION MAP

CLAIMS

  • No claims are published for this paper.

CONCEPTS

  • No concepts are published for this paper.

REFERENCES

Showing 1-45 of 45 references · Page 1 of 1

CITED BY

Showing 1-100 of 103 citing papers · Page 1 of 2

12
Next