Disaster medicine: analysis of research papers by Russian investigators based on artificial intelligence methods (2005–2021)

Relevance. Artificial intelligence is one of the fastest growing and promising technologies for processing and classifying natural text. The objective is to conduct a semantic analysis of domestic publications on emergency-related medical, biological and psychological problems, that are registered in the Russian Science Citation Index within 2005–2021 and where authors implement artificial intelligence methods and special software. Method. The object of the study is to analyze emergency-related medical, biological and psychological problems, published within 2005–2021 and registered in the Russian Science Citation Index. The analysis was carried out using VOSviewer program, as well as a neural network using the Python programming language. Results and discussion. For the purpose of semantic analysis, numerous scientific articles were imported from the Russian Science Citation Index into the VOSviewer program. Bibliometric visualization and clustering of key terms was conducted, to provide an insight into the publications and affiliations of main authors. In addition, cluster analysis and the ‘elbow’ method allowed to build a neural network in an automated mode to provide for more effective clustering; these data enabled us to identify 19 thematic clusters of scientific publications. Conclusion. In their publications investigators should be aware that the language of abstracts and keywords shall be inconformity with thesauruses, such as Medical Subject Headings (MeSH). These clarifications contribute to a more correct correlation between publications and existing headings. Further increase in the level of digitalization, including artificial intelligence methods, will allow to conduct more accurate research