Neural networks applied to retrocochlear diagnosis

Methodologies have been developed, based on insights from signal detection theory, to evaluate quantitatively the diagnostic performance of tests. Sev eral studies have demonstrated that, in fact, performance of a test buttery can be inferior to the best of the tests it includes. These studies have b een quite persuasive in damping enthusiasm for the test battery approach. B ecause the results of all tests in a battery were weighted equally in these studies, it is not surprising that an individual test with good sensitivit y and specificity is more effective diagnostically than a combination of te sts with poorer sensitivity and specificity. The authors of many of these s tudies were well aware of the limitations of this approach. In the present study neural networks were applied to evaluate audiological tests used to p redict retrocochlear pathology by differentially weighting the results of t he tests in the battery. This technique avoids some of the limitations of p revious approaches. Of the audiological tests evaluated in the present anal ysis, the superiority of the auditory brainstem evoked response (ABR) in pr edicting retrocochlear disease was again demonstrated. However, the results also demonstrated that identification accuracy could be improved by combin ing the ABR with other tests (in this case contralateral acoustic reflex at 2000 Hz, ipsilateral acoustic reflex at 2000 Hz, tone decay, and word reco gnition score). Further, it was demonstrated that performance could be impr oved over that obtained using dichotomous test measures (i.e., positive or negative presence of pathology) by using raw test measures in conjunction w ith ABR.