Abstract:
|
In medical diagnostic studies, a diagnostic test can be evaluated based on its sensitivity under a desired specificity. Existing methods for inference on this parameter include normal approximation-based approaches and empirical likelihood (EL) based approaches. These methods, however, have poor performance when the specificity is high, and some requires choosing smoothing parameters. We propose a new influence function based empirical likelihood method and a Bayesian empirical likelihood method to overcome such problems. Numerical studies are performed to compare the finite sample performance of the proposed approaches with existing methods. Our simulation results show that the proposed methods perform better in terms of both coverage accuracy and interval length.
|