Research on machine learning-based clinical prediction models: a bibliometric analysis

Area of Science:

• * Biomedical Informatics
• * Artificial Intelligence in Healthcare
• * Clinical Decision Support Systems

Background:

• * Machine learning (ML) is revolutionizing clinical prediction models (CPMs) using multidimensional patient data for enhanced risk stratification and decision-making.
• * The field of ML-CPMs has seen exponential growth, with nearly 250,000 publications by 2024, yet lacks comprehensive bibliometric analysis.
• * This study addresses the need for a systematic overview of the global research landscape in ML-CPMs.

Purpose of the Study:

• * To conduct a bibliometric and visualization analysis of global research on machine learning-based clinical prediction models (ML-CPMs).
• * To identify evolutionary trends, key research hotspots, and collaboration patterns within the ML-CPM field.
• * To provide insights into the current status and future directions of ML-CPMs.

Main Methods:

• * Retrieval of publications on ML-CPMs from Web of Science and Scopus databases (up to May 2025).
• * Utilization of bibliometric analysis tools including R, VOSviewer, and CiteSpace.
• * Generation of analyses on publication trends, collaboration networks, journal distributions, co-citation, clustering, and keyword co-occurrence.

Main Results:

• * Analysis of 8,619 publications from 118 countries, with exponential growth in publications since 2015 (R² = 0.9919).
• * China leads in publication volume, while the United States demonstrates the highest academic influence (H-index=105).
• * Key research hotspots include algorithm optimization, multimodal data integration, and model interpretability, with primary clinical applications in oncology, cardiovascular diseases, and critical care.

Conclusions:

• * ML-CPMs research exhibits significant global growth and international collaboration.
• * Persistent challenges include model interpretability, data heterogeneity, and privacy concerns.
• * Future research should emphasize external validation, clinical applicability, and human-AI collaboration for effective real-world implementation.