Analysis of Research Trends in Exercise Science Using LDA Algorithm: Focusing on the Period from 2015 to 2024

Abstract

PURPOSE: The purpose of this study was to analyze the research trends in exercise physiology over the past decade, ultimately providing new prospectives on the fields. METHODS: We targeted the titles of papers published on the official website of the Korean Society of Exercise Physiology between January 2015 and August 2024. In total, 367 articles were included. Among them, 98 articles had English titles and were analyzed after translation into Korean. The “Tokenization” process, which extracts noun words and “Stopwords” to eliminate unnecessary particles of articles, was executed. Subsequently, a common language was integrated, and terminology was standardized. The Latent Dirichlet Allocation (LDA) model was used to determine the optimal number of topics. RESULTS: The 367 articles analyzed were named Topics 1 to 8 (Topic 1: Exercise performance, Topic 2: Exercise rehabilitation, Topic 3: Resistance exercise, Topic 4: Covid/Exercise, Topic 5: Elderly/Exercise, Topic 6: Elite/Exercise, Topic 7: Physical assessment, and Topic 8: Metabolic syndrome). After dividing the years into four phases, the proportions of each topic were determined by phase. Among Topics 1-8, the topics that indicated a significant change were Topic 3 (rho=-0.104, p<.038), Topic 7 (rho=-0.507, p<.000), and Topic 8 (rho= 0.461, p<.000). CONCLUSIONS: Research trends over the last 10 years have shown that research topics have changed over time. The trend changes indicated a decrease in interest in physical assessments and a rapid increase in metabolic syndromes. This study highlights the need for health management studies in accordance with changes in the societal context, along with digital healthcare and wearable technology. © 2024 Korean Society of Exercise Physiology.