Digital twin (DT) technology, combined with advanced computational methodologies, represents a paradigm shift in sports science. DTs generate virtual athlete replicas through real-time data integration and predictive analytics. While computational capabilities and data acquisition have advanced rapidly, DT applications in sports remain fragmented, warranting systematic synthesis. This narrative review examines the applications of DT in sports, with a focus on big data analytics and deep learning for performance enhancement, injury prevention, and rehabilitation. A narrative review was conducted using publications from 2018 to 2025 across multiple databases. Eligible studies applied DT frameworks, Artificial Intelligence (AI), including machine learning algorithms and deep neural networks, or big data analytics in elite and amateur sport contexts, with a focus on football as a case study. Extracted data focused on technological approaches, clinical outcomes, and practical applications. DT applications cluster into three domains: (1) performance enhancement via biomechanical modelling (convolutional or recurrent neural networks); (2) injury prevention using ensemble learning and predictive risk models; and (3) rehabilitation optimization through multimodal sensors and virtual reality. Key examples include cycling telemetry, computer vision for technique correction, and real-time musculoskeletal monitoring. The integration of generative AI and the Internet of Things further enhances predictive accuracy and decision-making. DTs offer significant potential for proactive athlete management. Widespread adoption requires standardized protocols, clinical validation, and robust ethical frameworks for data privacy. Successful integration supports data-driven training, individualized recovery, and enhanced athlete welfare.