Logit-transformation Based Confidence Intervals for the Sensitivity of a Continuous-scale Diagnostic Test

Abstract

In the field of modern medicine, diagnostic tests play an important role in detecting diseased patients. The diagnostic accuracy of a test is measured by its sensitivity and specificity. In this paper, we focus on the interval estimation for the sensitivity of a continuous-scale diagnostic test at a fixed level of its specificity. We propose two logit-transformation based confidence intervals for the sensitivity of a test and compare them with two of the best existing intervals (BTI and BTII) proposed by Zhou and Qin. Our simulation studies show that the newly proposed logit- transformation based intervals (LTI and LTII) have shorter interval lengths than the existing BTI and BTII intervals. The applications of the new intervals are illustrated in two real examples.