Analysis of Research Trends and Hotspots of AI in Sports Training Based on BERTopic Modeling

Abstract

[Objective] With the rapid advancement of Artificial Intelligence (AI), there is a continuous emergence of hot technical topics concerning its application in the field of sports training. This study focuses on the application of AI in sports training, aiming to uncover the prominent technical topics within this domain and their evolutionary pathways.[Methods] By analyzing literature data from the Web of Science database (2015-2024) on AI applications in sports training, the BERTopic model extracted key topics. Subsequent evolutionary analysis summarized the research frontiers and hotspots of AI in sports training over the past decade.[Results/Conclusions] The BERTopic-based thematic modeling analysis revealed that research interest in AI for sports training was driven by dual factors: policy support and technological progress, reaching its peak publication volume in 2024. The research topics primarily coalesce into three primary clusters: 1) Clinical Rehabilitation, 2) Competitive and Educational Applications, and 3) Emerging Technologies. Clinical rehabilitation required solutions for data silos and standardization issues. Competitive scenarios needed deeper technological integration. Emerging directions like Generative AI, while still nascent, showed significant potential. Temporal evolution analysis revealed a trend toward integrating traditional technologies systematically, with cross-disciplinary collaboration propelling progress in motion recognition. Future research should prioritize multimodal data fusion, end-to-end system development, and cross-disciplinary cooperation to address bottlenecks like limited dynamic feedback and talent shortages, enabling AI to evolve from a tool-based assistant to an intelligent decision-maker. This study offers theoretical insights and practical guidance for understanding field frontiers, optimizing resource allocation, and accelerating technological industrializa.