Application of artificial intelligence in acute myocardial infarction: A bibliometric analysis from 2016 to 2025.

Abstract

Growing interest suggests that the broad application of artificial intelligence (AI) in acute myocardial infarction (AMI) has advanced diagnosis, prognosis, and treatment. AI methods applied to AMI research have recently yielded significant predictive outcomes. However, few studies have systematically analyzed this field using bibliometric approaches. We employ CiteSpace to examine the visual relationships of publications, identify key research trends and hotspots, and stimulate broader participation in AI-AMI studies. Data for this bibliometric analysis were retrieved from the Web of Science Core Collection. A systematic search strategy was developed using Medical Subject Headings to identify publications pertaining to AI in AMI from January 2016 to December 2025. The initial query returned 2089 records, which were subsequently refined through manual screening and algorithmic deduplication, yielding a final corpus of 2088 eligible publications. Bibliometric analyses were performed using CiteSpace (version 6.2.R6) with the following configuration: annual time slices of 1 year, a selection threshold of the top 25 most frequent items per slice (Top N = 25), and network pruning via the Pathfinder algorithm applied to both sliced and merged networks. Multidimensional analyses were conducted across six node types - country/region, institution, author, cited author, cited journal, and keyword - employing co-occurrence mapping, co-citation analysis, cluster detection, and burst term identification. These approaches collectively delineated the intellectual landscape, collaborative architectures, and thematic evolution characterizing the AI-AMI research domain. We analyzed publications, citations, and networks. Post-2016, publications surged from 24 (2016) to 517 (2025) with dramatically increasing citations. Global collaboration centered on the China (705 publications), USA (553 publications), and India (195 publications), with Harvard-linked institutions forming the dominant cluster (e.g., Harvard University: 100 publications, centrality = 0.12). Author networks identified key bridges Saba, Luca (13 publications, centrality = 0.05) and Pontone, Gianluca (centrality = 0.06), contrasting with high-output but low-centrality authors like Acharya, U Rajendra (21 publications). High-centrality cited authors (e.g., ACHARYA UR: centrality = 0.09) were pivotal nexuses. Journal analysis showed CIRCULATION (1375 cites) and J AM COLL CARDIOL (1120 cites) as citation leaders. Keywords revealed evolving focus: prevalent terms ("myocardial infarction": 698 freq., "machine learning": 603 freq.) were highly productive, whereas central terms ("prediction": centrality = 0.36, "acute myocardial infarction": centrality = 0.24) bridged domains. Keywords bursts shifted from computational themes pre-2020 (e.g., "neural network," strength = 8.31) to the practical application of AI post-2020 (e.g., "system," strength = 5.43 and "prevalence," strength = 4.61). AI-AMI research exhibits globally distributed multilevel collaboration structures anchored by major nations/institutions and critical author bridges. Knowledge dissemination hinges on domain-bridging journals like CIRCULATION. Thematic emphasis has evolved from computational techniques towards clinical applications, interconnected by high-centrality concepts.