Artificial Intelligence Diagnostic Accuracy and Clinical Utility in Allergic Rhinitis: Systematic Review

Dana A. Alrezq, Abdulaziz M. Aldaghmani, Jana J. Alali, et al.  Authorea. September 12, 2025.

DOI: 10.22541/au.175766646.65353803/v1

Abstract

Allergic rhinitis (AR) is one of the most common outpatient conditions, diagnosed through operator-dependent and resource-intensive methodologies. Consequently, there has been a surge in the development of diagnostic strategies using artificial intelligence (AI). This review aims to explore how AI can enhance diagnostic accuracy, personalize treatment, and support clinical decisions in AR care. The review followed PRISMA guidelines and was registered in PROSPERO. PubMed, Cochrane Central, Embase, and Google Scholar were searched for studies published between 2002 and 2025. Out of 1,109 identified studies, eight studies met the inclusion criteria. Eight studies investigated the application of AI in AR care, including ensemble machine learning classifiers, Attention U-Net architecture, smartwatch-based systems, and the eXtreme Gradient Boosting (XGBoost) with Shapley Additive Explanations (SHAP) algorithm. The Attention U-Net model showed high performance in segmenting wheal and erythema from skin prick test (SPT) images, achieving diagnostic accuracy of 0.9986 and a specificity of 0.9995. A key finding from this review is that AI models can serve as complementary tools beside gold standard methods. However, clinical adaptation will require external validation, interoperability with electronic health records, clinician friendly design, and adherence to Good Machine Learning Practice (GMLP). The lack of standardized comparator methods across studies remains a key limitation.

This is a preprint and has not been peer reviewed. Data may be preliminary.