Towards precision medicine for otology and neurotology: Machine learning applications and challenges

Hear Res. 2025 Nov 13;469:109473. doi: 10.1016/j.heares.2025.109473. Online ahead of print.

ABSTRACT

Advances in artificial intelligence, particularly machine learning and deep learning, in conjunction with the rise of personalised medicine, can facilitate tailored decision-making for diagnoses, prognoses, and treatment responses based on individual patient data. The multifaceted nature of symptoms and disorders in (neuro)otology, with their diverse aetiologies and subjective characteristics, makes this field an ideal candidate for computational personalised medicine. This narrative review critically synthesises applications of machine learning and deep learning in otology and neurotology published between 2013 and 2025. Relevant studies were identified through targeted searches of PubMed, Scopus, and Google Scholar using combinations of terms related to artificial intelligence, tinnitus, cochlear implants, and otologic or neurotologic disorders. Only peer-reviewed articles focusing on human applications of machine learning or deep learning in these fields were included, excluding theoretical papers or animal studies. Recent breakthroughs, such as the Whisper speech recognition model for cochlear implant simulations and large language models for refining tinnitus subgroup identification and therapy predictions, underscore the transformative potential of AI in improving clinical outcomes. This review is distinct in its emphasis on these emerging technologies and their integration into multimodal datasets, combining imaging, audiometric data, and patient-reported outcomes to refine diagnosis and treatment approaches. However, challenges including the lack of standardisation, limited generalisability of models, and the need for improved frameworks for multimodal data integration impede rigorous and reproducible implementation, topics that are critically explored in this review. Here, we explore the applications of machine learning, deep learning, and large language models in tinnitus, cochlear implants, and (neuro)tology, providing a critical analysis of recent advancements, persistent challenges, and recommendations for future research. By addressing these challenges and implementing recommended strategies, this review outlines a pathway for integrating cutting-edge artificial intelligence tools into clinical practice, underscoring their immense potential to revolutionise precision medicine in otology and neurotology and improve patient outcomes.

PMID:41274259 | DOI:10.1016/j.heares.2025.109473