Pathophysiology, Etiology, and Uses of AI in Otological Disorders Like Audiological Disorders, Ear Infections (AIED), Excess Wax, Tinnitus, Ménière’s Disease

Abstract

Otological diseases are a heterogeneous group with multiple etiologies and pathophysiologies which can impact on both auditory and vestibular function. The present study aimed at analysis of clinical, etiological and audiological characteristics of audiological disorders, autoimmune inner ear disease (AIED), excessive cerumen, tinnitus and Ménière's disease, and the possible use of the artificial intelligence (AI) in the diagnostic classification of these disorders. This study used a retrospective analytical design with a hypothetical sample of 250 patients which consisted of 50 patients from each of the five diagnostic groups. Analysis of clinical records, otoscopic findings, pure tone audiometry, speech audiometry and relevant findings of the vestibular function were performed. The pure-tone hearing thresholds varied significantly between the diagnostic groups (p<0.001) with AIED having the worst hearing loss. Recurrent vertigo, age, autoimmune/inflammatory history and hearing loss were all significant predictors of hearing impairment. Random forest, support vector machine and logistic regression were the evaluated AI methods with the highest diagnostic performance, with random forest having an accuracy of 91.6% and an AUC-ROC of 0.947. The results suggest that integrating clinical and audiological variables with AI analysis could improve the ability to distinguish among otological disorders and aid in clinical decision-making. These results need to be confirmed in large, multicentre clinical datasets before they can be routinely used in clinical practice.