Standard and Computed Tomography in the Evaluation of Neoplasms of the Chest

Inouye, Sharon K.

doi:10.7326/0003-4819-105-6-906

Cited by 23 publications

(11 citation statements)

References 115 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the Introduction, data published by Inouye and Sox (1986) were described. The data comprise the number of true positive diagnoses Y1 out of N1 diseased individuals and the number of false positive diagnoses YO out of NO nondiseased individuals for each of 14 studies.…”

Section: Application To Meta-analysismentioning

confidence: 99%

“…However, the fact that the ROC curve is unaffected by monotone transformation of Z is a complicating factor only recently solved by Metz, Herman, and Shen (199813). Nonparametric estimation is proposed by Lloyd (1998), while Moise et al (1985) and Weiand et al (1989), describe methods Inouye and Sox (1986). Summary ROC curves using the method of Moses et al (1993) …”

Section: Introductionmentioning

confidence: 98%

“…As an illustration, Figure 1 displays information on 14 independent studies of a method of detecting mediastinal metastases in patients with non-small-cell lung cancer from computer tomography, originally published by Inouye and Sox (1986). For each study, differing numbers of patients of known disease status were assessed by the method and then the empirical TPF and FPF were calculated.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Regression Models for Convex ROC Curves

Lloyd

2000

Biometrics

View full text Add to dashboard Cite

The performance of a diagnostic test is summarized by its receiver operating characteristic (ROC) curve. Under quite natural assumptions about the latent variable underlying the test, the ROC curve is convex. Empirical data on a test's performance often comes in the form of observed true positive and false positive relative frequencies under varying conditions. This paper describes a family of regression models for analyzing such data. The underlying ROC curves are specified by a quality parameter delta and a shape parameter mu and are guaranteed to be convex provided delta > 1. Both the position along the ROC curve and the quality parameter delta are modeled linearly with covariates at the level of the individual. The shape parameter mu enters the model through the link functions log(p mu) - log(1 - p mu) of a binomial regression and is estimated either by search or from an appropriate constructed variate. One simple application is to the meta-analysis of independent studies of the same diagnostic test, illustrated on some data of Moses, Shapiro, and Littenberg (1993). A second application, to so-called vigilance data, is given, where ROC curves differ across subjects and modeling of the position along the ROC curve is of primary interest.

show abstract

Section: Application To Meta-analysismentioning

confidence: 99%

Section: Introductionmentioning

confidence: 98%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Regression Models for Convex ROC Curves

Lloyd

2000

Biometrics

View full text Add to dashboard Cite

show abstract

“…In panel A, the pretest probability is 0.1, and the sensitivity and specificity are 0.74 and 0.79, respectively. 5 Recall that the sensitivity of a test represents the probability that a diseased patient will have a positive test; specificity denotes the probability that a nondiseased patient will have a negative test. The posttest probability, calculated with Bayes' theorem, is 0.28, as shown in the figure.…”

Section: Important Conceptual Lessons From Bayes' Theoremmentioning

confidence: 99%

“…For squamous carcinoma, the prevalence of mediastinal metastases is about 0.25, on average. 5 One can then adjust the initial estimate based on factors particular to the patient. In our example, we would decrease the probability of mediastinal metastases both because the nodule is small and because it is peripheral.…”

Section: Pretest Probabilitiesmentioning

confidence: 99%

Strategies for Test Selection in the Staging of Lung Neoplasms

Owens¹,

Sox²,

Inouye³

1989

Semin Respir Crit Care Med

View full text Add to dashboard Cite

Although several strategies have been proposed for the staging of lung neoplasms, there is still considerable disagreement as to the appropriate use of mediastinoscopy, computed tomography (CT), standard tomography (ST), and magnetic resonance imaging (MRI). We believe the disagreement is in part related to two problems. First, data on test performance have been inaccurate, inconsistent, or unavailable. Second, the implications of pretest probability and test performance on post-test probability have not been explicitly considered when test selection recommendations have been formulated.In this article we first discuss general concepts of test interpretation derived from Bayes' theorem. We then review test performance data for CT, ST, and mediastinoscopy and present general guidelines for test selection that can be applied to any proposed strategy for lung cancer staging. IMPORTANT CONCEPTUAL LESSONS FROM BAYES' THEOREMOf the possible tests one can consider for staging or diagnosis of lung cancer, all are imperfect. That is, both false-positive and false-negative test results occur. Even mediastinoscopy has a significant falsenegative rate. 1-4 Thus, one cannot know the patient's true state with certainty; intelligent test interpretation requires the use of probability.Bayes' theorem is a straightforward algebraic expression that allows calculation of the post-test probability of disease if the pretest probability, test sensitivity and specificity are known (see Appendix). A number of important lessons follow from an understanding of the rule:1. The pretest probability of disease has a dramatic effect on the post-test probability. Figure 1 illustrates the post-test probability of mediastinal metastases after a positive CT scan. In panel A, the pretest probability is 0.1, and the sensitivity and specificity are 0.74 and 0.79, respectively. 5 Recall that the sensitivity of a test represents the probability that a diseased patient will have a positive test; specificity denotes the probability that a nondiseased patient will have a negative test. The posttest probability, calculated with Bayes' theorem, is 0.28, as shown in the figure. In panel B, the calculation is repeated using identical sensitivity and specificity data, but with a pretest probability of 0.5. Note the post-test probability is now 0.78, considerably higher than before. In each case, the CT scan is "positive," but the post-test probability of disease is very different in the two situations. This example illustrates the importance of pretest probability: rational test interpretation is virtually impossible if one cannot estimate, at least roughly, the pretest probability. 2. The sensitivity (also called the true-positive rate) of a test primarily affects the interpretation of a negative result. In Figure 2 panel A shows the post-test probability of disease after positive and negative CT scans, calculated for a pretest probability of 0.50. The sensitivity and specificity are each 0.75. Sensitivity is defined as the probability that a dis-

show abstract

Statistical Methods for Meta‐Analysis

Zhou

Obuchowski

McClish

2002

Statistical Methods in Diagnostic Medicine

View full text Add to dashboard Cite

I would like to express the greatest 'thank you' to my advisor Dr. Haitao Chu for his continuous support in my four-year PhD study. I feel very fortunate to have met Haitao when I entered the University and to have worked with him since then. Beyond statistical, epidemiological, and clinical knowledge, I am glad that Haitao has spent a lot of time on training my critical thinking, which has been essential for me to become an independent biostatistics researcher. Also, Haitao has given me many opportunities to review other researchers' work as a referee and get involved in preparing grant proposals.These experiences have been greatly valuable for my long-term career.Many thanks also go to Drs. Bradley Carlin, James Hodges, and Gongjun Xu, who kindly served on my dissertation committee. They provided numerous helpful suggestions on my scientific writing as well as research topics. Moreover, I worked with Dr. Wei Pan as a research assistant for one year; his guidance let me get involved in genetic research and high-dimensional data analysis. In addition, several grants from National Institutes of Health and a Doctoral Dissertation Fellowship from the Graduate School financially supported my research in the last four years. They also provided funds to present my work in various scientific conferences. Last but not least, I would like to thank my family, especially my parents, for their love throughout my life; they are indispensable pieces for my happiness.

show abstract

Standard and Computed Tomography in the Evaluation of Neoplasms of the Chest

Cited by 23 publications

References 115 publications

Regression Models for Convex ROC Curves

Regression Models for Convex ROC Curves

Strategies for Test Selection in the Staging of Lung Neoplasms

Statistical Methods for Meta‐Analysis

Contact Info

Product

Resources

About