A Hearing-Model-Based Active-Learning Test for the Determination of Dead Regions.

The study tested only 14 ears from seven male subjects and one female subject, and all test ears had previously been selected for extensive basal dead regions. The reference standard was imperfect: Fast PTCs were irregular, and for ears 7L, 8L, and 8R no clear Fast PTC minimum was available, so the Smart DRT mean was used as the reference. One ear, 8R, did not agree with other methods and showed markedly different responses to the initial grid across runs. Two ears with tinnitus had higher false-positive rates, and ear 1R needed 63, 50, and 45 trials across runs to reach the same accuracy, compared with the average of 33 trials. The algorithm assumed a 3 dB psychometric-function standard deviation, but estimated standard deviations were about 10 dB for ears 1R, 8L, and 8R. The active-learning audiogram threshold was 3.2 dB below the standard audiometer threshold on average, with a standard deviation of the difference of 6.7 dB. The Discussion says the method may be useful when fitting a hearing aid, but the supplied results report estimates of the dead-region edge and agreement with other tests, not hearing aid fitting decisions or patient outcomes.

The only funding information supplied is that the funder was the Engineering and Physical Sciences Research Council.

Do not conclude that Smart DRT improves hearing aid fitting or patient outcomes. The study estimated the dead-region edge in 14 ears and agreed with other test methods in 13 of 14 cases, but it did not test fitting decisions or patient outcomes.