Artificial intelligence (AI) has demonstrated high accuracy in identifying the causes of seasonality. allergic rhinitis (SAR) offers a potential step forward in personalized allergy care.
Although SAR is typically caused by exposure to pollen, it remains difficult to diagnose accurately, especially in temperate regions where patients are often sensitized to multiple overlapping allergens. Precise identification of the causative allergen is essential to guide allergen-specific immunotherapy (AIT), the only disease-modifying therapy available. However, traditional diagnostic approaches rely heavily on clinician interpretation, which can be complex and variable.
In the @IT-2020 project, researchers developed a modular clinical decision support system (CDSS) powered by machine learning to improve the etiological diagnosis of SAR. This system integrated three progressive diagnostic layers: clinical history with skin puncture testing, molecular immunoglobulin E testing, and electronic clinical and environmental diary. These components were used to train an AI model against a “gold standard” diagnosis defined by experts.
AI diagnosis of seasonal allergic rhinitis shows high accuracy
The resulting model demonstrated excellent diagnostic performance, achieving an area under the receiver operating characteristic curve of more than 95%. Of note, this system maintained high accuracy across different patient populations, including external validation in separate cohorts from different geographic regions.
AI-driven CDSS also showed high adaptability. Its performance varied appropriately depending on patient complexity, and interpretability analysis confirmed that both clinical features and sensitization patterns meaningfully contribute to diagnostic decisions. Importantly, this system reproduces expert AIT prescriptions with considerable reliability, suggesting real-world clinical relevance.
AI diagnosis outperforms clinicians
In a direct comparison, the AI model outperformed 24 clinicians when diagnosing some patients, highlighting its potential for enhanced decision-making in allergy practices. Additionally, reducing the monitoring period to 45 days did not significantly compromise accuracy, suggesting that this approach may be practical in routine care settings.
Despite these promising findings, this study was proof of concept and relied on a relatively small cohort. Further independent validation and prospective trials are needed to determine whether AI-based diagnostics can improve long-term disease management and patient outcomes.
If confirmed, this approach has the potential to transform SAR management by enabling more accurate, efficient and standardized diagnosis, ultimately supporting personalized treatment strategies and improving the quality of life for patients with allergic diseases.
reference
Matricardi PM et al. Etiological diagnosis of seasonal allergic rhinitis supported by artificial intelligence: @IT-2020 project. J Allergy Clinical Immunol. 2026;DOI:10.1016/j.jaci.2026.03.011.
Featured image: Da from Adobe Stock
