2024•Hospital PediatricsRequires access

Receiver Operating Characteristic (ROC) Curves: The Basics and Beyond

Pearl W. Chang, Thomas B. Newman

Open publisher page 28 citations

Abstract

Diagnostic tests and clinical prediction rules are frequently used to help estimate the probability of a disease or outcome. How well a test or rule distinguishes between disease or no disease (discrimination) can be measured by plotting a receiver operating characteristic (ROC) curve and calculating the area under it (AUROC). In this paper, we review the features of ROC curves and interpretation of ROC curves and AUROC values. We highlight 5 underappreciated features of ROC curves: (1) the slope of the ROC curve over a test result interval is the likelihood ratio for that interval; (2) the optimal cutoff for calling a test positive depends not only on the shape of the ROC curve, but also on the pretest probability of disease and relative harms of false-positive and false-negative results; (3) the AUROC measures discrimination only, not the accuracy of the predicted probabilities; (4) the AUROC is not a good measure of discrimination if the slope of the ROC curve is not consistently decreasing; and (5) the AUROC can be increased by including a large number of people correctly identified as being at very low risk for the outcome of interest. We illustrate this last concept using 3 published studies.

About this research paper

What this paper is about

Diagnostic tests and clinical prediction rules are frequently used to help estimate the probability of a disease or outcome. How well a test or rule distinguishes between disease or no disease (discrimination) can be measured by plotting a receiver operating characteristic (ROC) curve and calculating the area under it (AUROC). In this paper, we review the features of ROC curves and interpretation of ROC curves and AUROC values. We highlight 5 underappreciated features of ROC curves: (1) the slope of the ROC curve over a test result interval is the likelihood ratio for that interval; (2) the optimal cutoff for calling a test positive depends not only on the shape of the ROC curve, but also on the pretest probability of disease and relative harms of false-positive and false-negative results; (3) the AUROC measures discrimination only, not the accuracy of the predicted probabilities; (4) the AUROC is not a good measure of discrimination if the slope of the ROC curve is not consistently decreasing; and (5) the AUROC can be increased by including a large number of people correctly identified as being at very low risk for the outcome of interest. We illustrate this last concept using 3 published studies.

Why it matters

OpenAlex reports 28 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

Diagnostic tests and clinical prediction rules are frequently used to help estimate the probability of a disease or outcome. How well a test or rule distinguishes between disease or no disease (discrimination) can be measured by plotting a receiver operating characteristic (ROC) curve and calculating the area under it (AUROC). In this paper, we review the features of ROC curves and interpretation of ROC curves and AUROC values. We highlight 5 underappreciated features of ROC curves: (1) the slope of the ROC curve over a test result interval is the likelihood ratio for that interval; (2) the optimal cutoff for calling a test positive depends not only on the shape of the ROC curve, but also on the pretest probability of disease and relative harms of false-positive and false-negative results; (3) the AUROC measures discrimination only, not the accuracy of the predicted probabilities; (4) the AUROC is not a good measure of discrimination if the slope of the ROC curve is not consistently decreasing; and (5) the AUROC can be increased by including a large number of people correctly identified as being at very low risk for the outcome of interest. We illustrate this last concept using 3 published studies.

Key concepts: Receiver operating characteristic, Cutoff, Medicine, Statistics, Confidence interval, Area under the curve, Interval (graph theory), Range (aeronautics)

Related papers

Back to paper searchBrowse research topicsOriginal source
Receiver Operating Characteristic (ROC) Curves: The Basics and Beyond — Research Paper | ScholarLens