Development of hearing screening application software based on speech-in-noise hearing test

Introduction.

Age-related hearing loss often goes undetected because it has few subjective symptoms, and people typically do not seek medical attention. Therefore, an easy method for screening hearing loss is needed. We developed a hearing screening application based on a speech-in-noise test. We compared its results with those of standard hearing tests and questionnaires to validate the application and determine the cut-off value for correct responses indicating hearing loss.

Methods.

The study included 150 outpatients, aged 20 years or older, with suspected hearing loss who underwent pure-tone audiometry at the Department of Otolaryngology-Head and Neck Surgery, Tohoku University Hospital. We performed pure-tone audiometry, speech-in-noise testing using an application, the Hearing Handicap Inventory for Adults, and another hearing questionnaire. In the speech-in-noise test, 10 monosyllables were presented in three different noise conditions. We compared each test in two groups: one with hearing loss (average pure-tone hearing threshold of 25 dB or more) and normal hearing (average pure-tone hearing threshold of less than 25 dB). Using the application, we drew an ROC curve with the hearing loss group as positive to compare the number of correct responses in speech-in-noise testing. We defined the cut-off value for the number of correct responses in the possible hearing loss group.

Results.

Correct responses in the speech-in-noise test using the application correlated significantly with a pure-tone hearing threshold (r=-0.56, P<0.0001). Significant differences in pure-tone hearing threshold, speech-in-noise testing, and two subjective hearing questionnaires were observed in the hearing loss group compared to the normal hearing group. The cut-off value was 18 items (60.0%), with a sensitivity of 0.63, specificity of 0.90, and AUC of 0.82.

Conclusion.

The speech-in-noise test developed with this application exhibited slightly lower sensitivity as a screening tool. However, our results indicated that approximate hearing ability could be inferred from the number of correct responses on the speech-in-noise test, and the high specificity suggested that the application may be useful for definitive diagnosis of hearing loss.