MULTIMODAL MACHINE LEARNING FOR DIFFERENTIAL DIAGNOSIS OF CHRONIC OTALGIA

Abstract

Chronic otalgia is a persistent ear pain condition that may arise from primary otologic disorders or referred pain from dental, temporomandibular, neurological, cervical, or head and neck pathologies. Because its causes are often overlapping and clinically complex, accurate differential diagnosis can be challenging when audiological, imaging, and clinical findings are interpreted separately. This study proposes a chronic multimodal machine learning framework for the differential diagnosis of chronic otalgia using integrated audiological, radiological, and clinical features. Audiological variables may include pure-tone audiometry, tympanometry, speech discrimination scores, and hearing asymmetry, while imaging features may be extracted from computed tomography or magnetic resonance imaging reports. Clinical variables may include pain duration, laterality, otoscopic findings, associated tinnitus, vertigo, dental history, temporomandibular joint symptoms, neurological signs, and prior treatment response. By combining these heterogeneous data sources, the proposed model aims to identify diagnostic patterns that may be missed through single-modality assessment. The multimodal approach is expected to improve classification performance, support early recognition of referred otalgia, and assist clinicians in prioritizing further investigations. The study highlights the potential of machine learning as a clinical decision-support tool for chronic otalgia, particularly in cases where conventional evaluation produces uncertain or overlapping diagnostic impressions. Overall, the framework may contribute to more accurate diagnosis, reduced diagnostic delay, and personalized management of patients presenting with chronic ear pain.