Global research trends on artificial intelligence-based diagnostic approaches in allergy: A bibliometric and scientometric study
Tuğba Songül Tat, Seda AltınerBackground
Artificial intelligence (AI) has emerged as a promising tool to address the diagnostic evaluation of allergic diseases. This study aims to assess global research trends and scientific collaborations in AI-based diagnostic approaches for allergic diseases.
Methods
This bibliometric and scientometric study utilized the Web of Science Core Collection. A comprehensive topic search covering allergic diseases, diagnostic terminology, and AI-related keywords was conducted (1993–2025). Only original research articles were included (n = 1,277). Descriptive analyses, poisson regression, and citation metrics (total citation, average citations per year) were performed. VOSviewer was employed to construct networks of co-authorship, country collaboration, citation, co-citation, and keyword co-occurrence. AI methods were categorized using an operational coding scheme of nine mutually exclusive categories, with independent dual review.
Results
1,277 articles authored by 7,489 researchers from 95 countries were analyzed. Publication output rose sharply after 2018, with 80.3% of all articles published between 2019 and 2025. Machine learning (ML) dominated the methodological landscape (58.5%), followed by deep learning (10%). Keyword mapping identified “machine learning,” “asthma,” “deep learning,” and “prediction” as central themes. Co-authorship and collaboration networks showed the USA, China, England, and Germany as the most influential contributors. Highly cited papers were predominantly ML-based. Journal analysis showed that Scientific Reports and PLOS One were the most productive sources, while the Journal of Allergy and Clinical Immunology (JACI) and JACI–In Practice were the central and most influential citation hubs.
Conclusions
This study provides a comprehensive scientometric overview of AI-based approaches contributing to allergy-related diagnostic research. The results show an increase in publications and methodological variety, especially in recent years.