Applications of artificial intelligence in cerebral palsy research: A bibliometric analysis
Lin Gao, Chenggui Xu, Shouyao Zhang, Pengbin Zhao, Exian Shan, Xiantao TaiObjective
Research on the application of artificial intelligence (AI) in the diagnosis and treatment of cerebral palsy (CP) has grown steadily. However, the overall research landscape, collaboration patterns, and thematic evolution in this field have not yet been systematically summarized. This study aims to clarify the current research status, shifts in research hotspots, and existing limitations through bibliometric analysis, thereby providing a comprehensive reference for future investigations.
Methods
A total of 278 English publications concerning AI applications in CP over the past 20 years were included in the analysis. Multiple bibliometric tools, including R-bibliometrix, VOSviewer, Scimago Graphica, and Pajek, were used to systematically analyze publication distribution, collaboration networks, intellectual base, and the temporal evolution of research hotspots in this field.
Results
This field demonstrates rapid growth and interdisciplinary development. Global research output is dominated by European and North American countries, characterized by close intraregional collaboration, whereas in-depth cross-team, interdisciplinary, and transnational joint research remains scarce. Two core research themes have emerged: early screening and risk prediction for CP in high-risk infants, and intelligent motor function assessment for children with confirmed CP. Research hotspots have gradually shifted from traditional machine learning to deep learning-based automatic feature extraction. At present, the field suffers from imbalanced research priorities, insufficient model interpretability, and limited clinical translation. Inconsistent data standards and overreliance on unimodal data greatly restrict the generalizability and clinical applicability of intelligent models.
Conclusion
Future research should foster in-depth multi-center, interdisciplinary, and transnational scientific collaborations, build standardized multimodal datasets, and develop interpretable intelligent models. Meanwhile, understudied directions such as rehabilitation response prediction and long-term prognosis tracking should be explored. Strengthening the clinical translation of algorithmic models will further refine the interdisciplinary research system of AI-assisted precision rehabilitation for CP, and promote the high-quality and intelligent development of pediatric rehabilitation.