Mapping the evolution of artificial intelligence applications in bone metastasis: a bibliometric and visualization analysis

Bone metastasis represents a major complication of advanced malignancies and is associated with substantial skeletal morbidity, impaired quality of life, and complex clinical decision-making. Artificial intelligence (AI) has rapidly emerged as an important computational approach for analyzing heterogeneous biomedical data. The developmental landscape, intellectual evolution, and future clinical directions of artificial intelligence applications in bone metastasis remain insufficiently characterized.

A comprehensive bibliometric analysis was performed using publications retrieved from the Web of Science Core Collection, PubMed, and Scopus from January 2006 to June 2026. Bibliometric indicators, including annual publication trends, journal distribution, countries, institutions, authors, citation patterns, keyword co-occurrence, thematic evolution, and reference bursts, were analyzed using CiteSpace, VOSviewer, and R-based bibliometric tools.

A total of 540 publications involving 3,439 authors from 237 journals were identified. Research output increased substantially after 2020, with an annual growth rate of 24.01%, reflecting rapid expansion of AI-related investigations in bone metastasis. China contributed the largest number of publications, whereas the United States, Japan, and European countries showed substantial scientific influence and international collaboration. Thematic analysis demonstrated a progressive transition from early computer-aided diagnosis and bone scintigraphy analysis toward deep learning, radiomics, multimodal prediction, and language-based artificial intelligence applications. Imaging-related approaches remained the dominant research direction, while emerging studies increasingly focused on clinical information extraction, prognostic modeling, and integration of heterogeneous data sources. A small number of recent studies explored language-based AI for clinical information extraction, data organization, and knowledge support, suggesting an emerging but still limited extension beyond conventional imaging applications.

AI research in bone metastasis has evolved from automated lesion recognition toward broader applications in disease characterization, risk assessment, and clinical decision support. Imaging, machine learning, deep learning, and radiomics remain the principal foundations of the field, whereas language-based AI currently represents a comparatively small and emerging research direction. Future studies should prioritize multicenter validation, multimodal data integration, prospective evaluation, and clinically meaningful endpoints. Large language model systems may warrant further investigation as potential components of future disease-specific AI frameworks, although their clinical utility in bone metastasis remains to be established.