Regional confidence bands for ROC curves

Jensen, Katrin; Müller, Hans‐Helge; Schäfer, Helmut

doi:10.1002/(sici)1097-0258(20000229)19:4<493::aid-sim352>3.0.co;2-w

Cited by 32 publications

(41 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Receiver operating characteristic (ROC) curves were created for selected variables, to evaluate the sensitivity and specificity of PFT parameters as predictors for percent-of-predicted peak V O 2 . 13 The Youden index was calculated to find the cutoffs for variables. The Youden index is defined as the maximum vertical distance between the ROC curve and the diagonal or chance line, and is calculated as maximum (sensitivity ϩ specificity Ϫ1).…”

Section: What This Paper Contributes To Our Knowledgementioning

confidence: 99%

Inspiratory Fraction Correlates With Exercise Capacity in Patients With Stable Moderate to Severe COPD

et al. 2013

View full text Add to dashboard Cite

BACKGROUND: Exercise intolerance is the hallmark of COPD. Static lung hyperinflation and increased dynamic hyperinflation during exercise are associated with reduced functional capacity in COPD patients. Inspiratory capacity correction for the total lung capacity, defined as inspiratory fraction (IF), may be functionally more representative than other traditional indices in these patients. OBJECTIVE: To investigate the association between IF and exercise capacity in patients with stable, moderate to severe COPD. METHODS: Fifty COPD subjects and 34 healthy volunteers constituted the study cohort. Pulmonary function and cardiopulmonary exercise testing were performed, and ventilation and gas exchange parameters were measured. RESULTS: IF was significantly correlated with percent-of-predicted peak oxygen consumption (V O 2 ) in the subjects with COPD (r ‫؍‬ 0.52, P < .001). IF was an independent predictor of reduced exercise capacity in the COPD subjects, and was more sensitive and specific than percent-of-predicted FEV 1 . Statistical analysis generated the equation: percent-of-predicted peak V O 2 ‫؍‬ 65.9 IF ؉ 0.45 percent-of-predicted FEV 1 ؉ 35.8 (R C 2 ‫؍‬ 0.39, P < .001). The subjects with IF < 0.23 had more severe lung hyperinflation and less exercise capacity than the subjects with IF > 0.23. At peak exercise, the breathing frequencies of the 2 groups were similar, whereas the low-IF subjects had reduced peak minute ventilation and peak tidal volume, relative to the high-IF subjects. CONCLUSIONS: Compared to FEV 1 , IF is a robust factor to reflect lung hyperinflation and to estimate the exercise capacity of subjects with stable moderate to severe COPD.

show abstract

Section: What This Paper Contributes To Our Knowledgementioning

confidence: 99%

Inspiratory Fraction Correlates With Exercise Capacity in Patients With Stable Moderate to Severe COPD

et al. 2013

View full text Add to dashboard Cite

show abstract

“…The vast majority of the ROC literature on confidence intervals investigates confidence intervals on the ROC curve itself, constructed in either a point-wise manner (Dukic & Gatsonis, 2003;McNeil & Hanley, 1984;Platt et al, 2000;Tilbury et al, 2000;Zou et al, 1997) or as a global confidence band (Dukic & Gatsonis, 2003;Jensen et al, 2000;Ma & Hall, 1993). Macskassy et al (2005) give a good review of this work accessible to a machine learning audience.…”

Section: Confidence Intervals On Cost Linesmentioning

confidence: 99%

Cost curves: An improved method for visualizing classifier performance

2006

View full text Add to dashboard Cite

This paper introduces cost curves, a graphical technique for visualizing the performance (error rate or expected cost) of 2-class classifiers over the full range of possible class distributions and misclassification costs. Cost curves are shown to be superior to ROC curves for visualizing classifier performance for most purposes. This is because they visually support several crucial types of performance assessment that cannot be done easily with ROC curves, such as showing confidence intervals on a classifier's performance, and visualizing the statistical significance of the difference in performance of two classifiers. A software tool supporting all the cost curve analysis described in this paper is available from the authors.

show abstract

“…Model discriminatory capacity and calibration were assessed by computing the area under the receiver operating characteristic curve and by using the Hosmer-Lemeshow goodness-of-fit test, respectively [9]. The concordance statistics (c-statistics) of each of the logistic regression models of 200 bootstrap sub-samples were used to compute a 95% CI for the area under the receiver operating characteristic curve [10].…”

Section: In-hospital Adverse Eventsmentioning

confidence: 99%