Academic Impact vs. Societal Attention: A Dual-Analysis of Top-Cited Artificial Intelligence Articles in Medicine

Objective This study aims to examine the relationship between the academic citation success and social visibility of Artificial Intelligence (AI)-based medical research using bibliometric and altmetric methodologies. Methods The top 100 most-cited articles indexed in the Web of Science Core Collection from 1 January 2023 to 27 January 2026 were analyzed; citation counts and Altmetric Attention Scores (AAS) were retrieved on 27 January 2026 (See Methods for the full search query). Academic impact was measured by Web of Science citation counts, while social impact was evaluated using the Altmetric Attention Score (AAS). Data were assessed through Spearman’s correlation analysis and the Mann-Whitney U test. Results A statistically significant but weak positive correlation was identified between citation counts and AAS (r = 0.299, p = 0.0025). Open access status characterized 92% of the articles. The highest academic impact was achieved by the ChatGPT-USMLE study by Kung et al. (2023) with 2,193 citations, whereas the highest social impact was held by the "Physician vs. Chatbot" study by Ayers et al. (2023) (AAS: 6,388). A notable finding was that publications originating from China exhibited remarkably low altmetric scores (Median AAS: 12) despite high academic citation rates, suggesting a 'digital isolation' effect that may stem from Western-centric altmetric data coverage. Conclusion Academic success and societal popularity are governed by distinct dynamics, indicating the need for researchers to adopt science communication strategies and for funding agencies to use multidimensional impact metrics. While academia prioritizes conceptual depth—such as ethics and methodology—the general public shows greater interest in sensational competition (e.g., physician vs. AI). It is recommended that researchers enhance their science communication competencies and that funding agencies adopt multidimensional evaluation approaches.