Abstract
Background: Immersive technologies such as virtual reality (VR), augmented reality (AR), mixed reality (MR), and extended reality (XR) are increasingly applied in patient care across multiple clinical domains. The rapid growth of this field demands efficient strategies for identifying and organizing the available literature. Artificial intelligence (AI)-powered tools have emerged as potential aids for literature retrieval, but their capacity to support rigorous evidence synthesis remains insufficiently evaluated. Objectives: This study aimed to describe the current state of clinical applications of immersive technologies in patient care based on literature published since 2023, and to compare the utility of two AI-assisted tools (OpenEvidence and ChatGPT) against a structured bibliographic search in Scopus for identifying, organizing, and contextualizing the available evidence. Methods: A hybrid literature retrieval strategy was employed. A structured search was conducted in Scopus using predefined terms related to immersive technologies and clinical applications, limited to articles and reviews published from 2023 onward in health-related subject areas. In parallel, the same topic was queried using OpenEvidence and ChatGPT. All AI-suggested references were manually verified for existence, publication year, and alignment with inclusion criteria. Articles were organized into predefined clinical application areas derived from two recent comprehensive reviews. The AI-retrieved articles were compared against the Scopus results in terms of thematic coverage, bibliometric visibility (citation count and ranking position), and recovery of highly cited articles (>50 citations). Results: The Scopus search identified a broad corpus of recent publications on immersive technologies in patient care. OpenEvidence retrieved 95 articles covering all 16 predefined clinical areas, while ChatGPT identified 22 articles but did not recover publications in three areas. Of the 56 articles with >50 citations in Scopus, only 6 (10.7%) were identified by either AI tool. The median ranking positions of AI-retrieved articles within the Scopus corpus were 511 for ChatGPT and 808 for OpenEvidence, indicating that most AI-suggested references were not among the highest-cited publications. The AI tools contributed to the identification of one additional thematic category (palliative and end-of-life care) not included in the initial classification. A final thematic map comprising 17 clinical application areas was established, with representative articles selected for each domain. Conclusions: In the queries analyzed, AI-assisted retrieval contributed to thematic mapping, while the two tools together identified 6 of the 56 publications with >50 citations in the Scopus bibliographic comparator (10.7%). OpenEvidence covered all 16 predefined clinical areas, whereas ChatGPT covered 13. Combining the retrieved publications with manual verification yielded a thematic map of 17 application areas. These findings describe the observed thematic and bibliometric coverage; they do not establish retrieval completeness or adequacy for clinical evidence synthesis.