Artificial Intelligence for Fracture Risk Prediction: A Bibliometric Analysis of Research Hotspots and Evolutionary Trends

Abstract

Artificial intelligence (AI) has emerged as a promising approach for improving fracture risk prediction. This study aimed to provide a comprehensive bibliometric overview of research on AI in fracture risk prediction by mapping its temporal evolution, collaborative structure, thematic framework, and disciplinary distribution. Publications were retrieved from the Web of Science Core Collection using a predefined search strategy covering January 2001 to December 2025, including only English-language articles and reviews. After screening and data cleaning, a total of 943 publications were analyzed. Bibliometric analyses were conducted using VOSviewer (v1.6.20) and the Bibliometrix R package (v4.5.1), examining annual publication trends, co-authorship networks at the author, institutional, and country levels, keyword co-occurrence clusters, and subject category distributions. The results showed that research output exhibited three developmental phases, with a marked acceleration after 2020. Collaboration networks demonstrated a clustered and moderately centralized structure, with core institutions and countries—particularly the United States and major European partners—occupying central positions. Keyword co-occurrence analysis identified four principal thematic clusters: AI-based predictive modeling, imaging-derived radiomics and opportunistic screening, FRAX-centered clinical risk assessment, and comorbidity- and pharmacotherapy-related applications. Subject category analysis further revealed progressive interdisciplinary integration, with increasing contributions from radiology and computer science alongside traditional clinical specialties. Overall, AI-based fracture risk prediction research has evolved from clinically grounded risk stratification toward computationally augmented, imaging-integrated, and translationally oriented predictive frameworks. The field demonstrates rapid growth, expanding international collaboration, and increasing interdisciplinary convergence, suggesting continued advancement toward real-world clinical implementation.

Ähnliche Arbeiten