목적: The advances in AI (Artificial Intelligence) that are transforming various fields have yet to applied in hearing. Traditional hearing tests, such as pure-tone audiometry (PTA), rely on a labour-intesive service process that fails to provide accessibility for early diagnosis hearing loss. In this perspective, the application of AI with hearing assessment brings about technological innovation to the hearing field. Thus, the present study purpose to develop a kiosk-based AI medical device (Kiosk Hearing) for hearing field. 방법:The Kiosk Hearing provide a self-hearing assessment based on kiosk and offer an early diagnosis for hearing loss based on AI. It consisted of a questionnaire, PTA, and an AI model for tympanic membrane (TM) analysis. The questionnaire was adopted RHHI-K (Revised Hearing Handicap Inventory for Korean), and reliability and validity had been completed. An automatic algorithm was developed for PTA, and the performance of the clinical audiometer was validated based on IEC 60645-1:2017. The AI model for TM analysis used deep learning techniques for ear diseases based on segmentation of the TM into five substructures. 결과:The number of four prototypes of the Kiosk Hearing have been developed with dedicated android SW, remote management services, and Medbiz server. The reliability and validity of the RHHI-K have been significantly verified, and the performance of the audiometer meet the requirements of IEC 60645-1. The ROC curve of the AI model for TM was obtained for optimal conditions for diagnosis. The highest area (0.911) was found for a combination of the malleus, cone of light, and umbo compared with the corresponding area under the curve of 0.737-0.873 for each substructure. 결론:The Kiosk Hearing can enhance accessibility to hearing assessment and ensure reliability through AI-based diagnosis. Moreover, it can facilitate timely and appropriate referral consultations, thereby improving the quality of life for hearing loss patient in the context of primary care.