A closed max-t test for multiple comparisons of areas under the ROC curve
Research output: Contribution to journal › Journal article › Research › peer-review
Standard
A closed max-t test for multiple comparisons of areas under the ROC curve. / Blanche, Paul; Dartigues, Jean-Francois; Riou, Jeremie.
In: Biometrics, Vol. 78, No. 1, 2022, p. 352-363.Research output: Contribution to journal › Journal article › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - JOUR
T1 - A closed max-t test for multiple comparisons of areas under the ROC curve
AU - Blanche, Paul
AU - Dartigues, Jean-Francois
AU - Riou, Jeremie
PY - 2022
Y1 - 2022
N2 - Comparing areas under the ROC curve (AUCs) is a popular approach to compare prognostic biomarkers. The aim of this paper is to present an efficient method to control the family-wise error rate when multiple comparisons are performed. We suggest to combine the max-t test and the closed testing procedures. We build on previous work on asymptotic results for ROC curves and on general multiple testing methods to efficiently take into account both the correlations between the test statistics and the logical constraints between the null hypotheses. The proposed method results in an uniformly more powerful procedure than both the single-step max-t test procedure and popular stepwise extensions of the Bonferroni procedure, such as Bonferroni-Holm. As demonstrated in this paper, the method can be applied in most usual contexts, including the time-dependent context with right censored data. We show how the method works in practice through a motivating example where we compare several psychometric scores to predict the t-year risk of Alzheimer's disease. The example illustrates several multiple testing settings and demonstrates the advantage of using the proposed methods over common alternatives. R code has been made available to facilitate the use of the methods by others.
AB - Comparing areas under the ROC curve (AUCs) is a popular approach to compare prognostic biomarkers. The aim of this paper is to present an efficient method to control the family-wise error rate when multiple comparisons are performed. We suggest to combine the max-t test and the closed testing procedures. We build on previous work on asymptotic results for ROC curves and on general multiple testing methods to efficiently take into account both the correlations between the test statistics and the logical constraints between the null hypotheses. The proposed method results in an uniformly more powerful procedure than both the single-step max-t test procedure and popular stepwise extensions of the Bonferroni procedure, such as Bonferroni-Holm. As demonstrated in this paper, the method can be applied in most usual contexts, including the time-dependent context with right censored data. We show how the method works in practice through a motivating example where we compare several psychometric scores to predict the t-year risk of Alzheimer's disease. The example illustrates several multiple testing settings and demonstrates the advantage of using the proposed methods over common alternatives. R code has been made available to facilitate the use of the methods by others.
KW - biomarker
KW - closed testing
KW - max‐
KW - t test
KW - multiple testing
KW - ROC curve
KW - survival analysis
KW - OPERATING CHARACTERISTIC CURVES
KW - GENERAL CONTRASTS
U2 - 10.1111/biom.13401
DO - 10.1111/biom.13401
M3 - Journal article
C2 - 33207001
VL - 78
SP - 352
EP - 363
JO - Biometrics
JF - Biometrics
SN - 0006-341X
IS - 1
ER -
ID: 253443962