Abstract
Introduction: In clinical medicine, different diagnostic tests may be used either to show the presence (ruling in) or absence (ruling out) of a certain disease. New statistical methods have been developed for evaluating the performance of diagnostic tests. The Kullback-Leibler (K-L) distance is a fundamental concept in information theory. This distance can be used for obtaining more detailed findings with respect to the performance of diagnostic tests when disease status has more than two classes.
Methods: K-L distance measures the distance between two probability distributions. The K-L distances, D(f//g) and D(g//f), measure the rule in and rule out potentials of a test, respectively. When comparing two or more tests, the test with the highest D(f//g) distance is the best test for ruling in, and the one wilh the highest D(g//f) is the best for ruling out the disease. In this study, the performances of Two Dimensional Echocardiography (TDE) and Tissue Doppler Echocardiography (TDEI) images parameters were evaluated for determining of etiology of pulmonary hypertension (PHT).
Results: Septum IVRT (ms), RV mid SR(1/s), RV apical SR(1/s), RVFAC (%) are the best variables in terms of ruling in and ruling out the disease for etiology of PHT.
Conclusion: Disease usually has more than two classes. When the disease status has more than two classes and the diagnostic variable is ordinal, the ruling in and ruling out performance of this variable can be evaluated by K-L distance.