Systematic review on machine learning applications in Paralympic sports: current practice and future research.

PURPOSE: This review explores the applications and challenges of machine learning (ML) in Paralympic and adaptive sports, focusing on performance profiling, injury prediction, movement analysis, and assistive technology development. MATERIALS AND METHODS: A comprehensive search of six electronic databases (PubMed, Web of Science, IEEE Xplore, Scopus, SPORTDiscus, and Google Scholar) was conducted, identifying 17 studies that applied ML in Paralympic and adaptive sports. Each study was evaluated for quality using the Prediction model Risk of Bias Assessment Tool (PROBAST). RESULTS: ML applications were found across four key areas: performance profiling, classification, movement analysis, and assistive technology. While advancements have been made in areas like para-swimming and wheelchair sports tracking, challenges remain due to small sample sizes, data heterogeneity, and the need for domain-specific adaptations to address the unique biomechanics and equipment used by athletes with disabilities. CONCLUSIONS: The application of ML in Paralympic sports has the potential to significantly enhance performance analysis and injury prevention. However, current models require further refinement to account for the diversity of impairments, equipment variations, and movement strategies in adaptive sports. Future research should focus on developing robust, interpretable ML models that are adaptable to the unique needs of Paralympic athletes. Improved validation practices and interdisciplinary collaboration are essential to ensure the translation of ML innovations into practical, real-world applications that support athletes' performance and well-being.