A Bibliometric Analysis of the Artificial Intelligence Application in Air Pollution (2007-2023): Evolution of Hotspots and Research Trends

Air pollution significantly threatens human health, leading to extensive research on its various parameters. In this study, we conducted a bibliometric analysis to assess the progress of artificial intelligence (AI) applications in air pollution research from 2007 to 2023. We utilized a comprehensive dataset from the Web of Science (WoS) core database and applied CiteSpace, bibliometrix of R package, VOSviewer, Alluvial Diagram Generator and Scimago Graphica to analyze countries, institutions, journals, authors, co-cited articles, keywords, and the dual-map overlay of journal categories. The results indicate a significant increase in publications since 2018, with China (979 publications) and the USA (459 publications) at the forefront. The Chinese Academy of Sciences tops the list with 106 papers and 3366 citations, and a centrality score of 0.22. Two authoritative journals, Atmospheric Environment and Science of the Total Environment, have the most citations with 3916 and 3501. However, research teams remain dispersed, with no large-scale collaborations formed to date. Keywords encompass "deep learning," "artificial neural network," "indoor air quality," "aerosol optical depth," "random forests," and "hospital admissions." The transformation of academic categories over the past 17 years, from seven initial groups in 2007 to the formation of ten new interdisciplinary groups between 2022 and 2023, underscores the broad impact of AI technologies and their expanding applications across various fields. The dual-map overlay analysis uncovers surprising connections between veterinary medicine and atmospheric research, highlighting the potential impact of air pollution on livestock and human health, emerging as a potential hotspot for future studies. Moreover, AI is demonstrating its interdisciplinary reach within the agricultural sector.