Artificial Intelligence in Deep Vein Thrombosis: A Bibliometric Analysis of Diagnostic and Predictive Research

Abstract

Background: Artificial Intelligence (AI) use has been growing in Deep Vein Thrombosis (DVT) and Venous Thromboembolism (VTE) studies for diagnosis, risk prediction, prognosis and clinical decision-making. Although the field has seen rising trends over the past few years, a thorough bibliometric analysis of the research in the field of AI-powered DVT is still lacking.
Objectives: The aim of this study was to profile the global scientific landscape of the applications of AI in DVT and VTE by analysing the publication patterns, prominent contributors, collaboration networks, and research themes.
Methods: A bibliometric analysis was performed on the literature retrieved from PubMed with a search made up to 18/05/2026. The structured Boolean search incorporating artificial intelligence, DVT/VTE and clinical terms identified 561 publications. VOSviewer (v1.6.20) and Biblioshiny (Bibliometrix R package) were used to analyze data. The bibliometric indicators used were scientific production, citation performance, co-authorship and institutional collaboration, country productivity, keyword co-occurrence, thematic evolution and citation mapping.
Results: Scientific production also rising significantly in recent years, As a result of the increasingly wide incorporation of machine learning and computational science in the study of thrombosis. The United States and the United Kingdom were the top contributors, and the University of Maryland system came out as a significant hub of collaboration. Commonly used terms were “machine learning,” “deep vein thrombosis,” “venous thromboembolism,” and “artificial intelligence.” Key themes that developed included deep learning, natural language processing, imaging-based AI, and prediction of COVID-19-associated thrombosis. Citation analysis revealed basic research on VTE risk scoring and predictive modelling. Conclusion: while AI-powered DVT research is on the rise, there is a need for further prospective validation, explainable AI, and anticoagulation-focused applications.