Artificial Intelligence in Rhinology: Current Applications, Clinical Utility, and Future Directions—A Systematic Review

Abstract

Background: Artificial intelligence (AI) has emerged as a powerful adjunct in rhinology, with expanding applications in diagnostic imaging, endoscopic assessment, disease classification, surgical planning, and clinical decision support. Recent advances in machine learning (ML), deep learning (DL), and convolutional neural networks (CNNs) have demonstrated promising diagnostic performance across a wide range of rhinologic disorders. However, the available evidence remains heterogeneous, and its readiness for routine clinical implementation requires critical evaluation.

Objective: To systematically review the current evidence on the applications of artificial intelligence in rhinology, with emphasis on its diagnostic performance, clinical utility, limitations, and future perspectives.

Methods: A systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. Electronic searches of PubMed/MEDLINE, Scopus, Embase, Web of Science, IEEE Xplore Digital Library, and the Cochrane Library were performed from database inception to May 31, 2025. Studies investigating AI-based techniques—including ML, DL, CNNs, natural language processing, and large language models—in rhinologic diseases were eligible for inclusion. Two reviewers independently screened studies, extracted data, and assessed methodological quality. Owing to substantial heterogeneity among studies, findings were synthesized narratively.

Results: Twenty-seven records were identified through database searching. Following duplicate removal and eligibility assessment, 12 studies published between 2019 and 2025 were included. AI applications encompassed computed tomography interpretation, endoscopic image analysis, chronic rhinosinusitis classification, nasal polyp detection, cytological assessment, radiomics, clinical decision support, surgical planning, and prediction of treatment outcomes. CNN- and DL-based models demonstrated high diagnostic performance in image-based applications, with reported diagnostic accuracies ranging from approximately 74% to 99% and area under the receiver operating characteristic curve (AUC) values of up to 0.93. AI-assisted systems improved diagnostic consistency, reduced interobserver variability, and supported clinical decision-making. Nevertheless, the available evidence was predominantly based on retrospective, single-center studies with relatively small datasets and limited external validation.

Conclusion: Artificial intelligence has significant potential to enhance rhinologic practice by improving diagnostic accuracy, optimizing image interpretation, facilitating surgical planning, and supporting personalized patient care. Despite encouraging results, current evidence is constrained by methodological heterogeneity and limited clinical validation. Large-scale multicenter prospective studies, standardized reporting frameworks, and externally validated, explainable AI models are essential before widespread integration of AI into routine rhinologic practice.