Background: Artificial intelligence (AI) is increasingly integrated into sports science, offering novel tools for injury surveillance, performance optimization, and psychological well-being monitoring in athletes. Despite growing adoption of machine learning (ML), computer vision, and wearable-sensor technologies, evidence regarding their clinical effectiveness and translational value remains fragmented across disciplines.

Objective: This article reviews and synthesizes current scientific evidence on the application of AI-based systems for reducing musculoskeletal injury risk and enhancing physical and psychological performance in athletic populations, with emphasis on methodological rigor and practical implementation.

Methods: A narrative synthesis was conducted drawing on systematic reviews, scoping reviews, and primary studies indexed in PubMed, Scopus, Web of Science, IEEE Xplore, and SPORTDiscus, focusing on peer-reviewed literature published between 2015 and 2026. Studies employing machine learning, deep learning, computer vision, wearable sensors, and AI-driven psychological monitoring in sport contexts were prioritized.

Results: Evidence indicates that AI-integrated wearable technologies and predictive analytics (e.g., random forests, convolutional neural networks, force-plate–derived biomechanical models) can identify injury risk factors, particularly in lower-limb musculoskeletal injuries, and support individualized rehabilitation protocols. Concurrently, AI-based facial expression analysis, physiological signal processing, and machine learning classifiers show emerging capacity to detect early markers of psychological stress and mental fatigue in elite athletes, complementing established sports-psychiatry frameworks. However, heterogeneity in study design, limited external validation, and "black-box" model interpretability remain significant barriers to clinical translation.

Conclusion: AI holds substantial promise for a proactive, individualized approach to athlete health management spanning biomechanical injury prevention and mental health surveillance. Future research should prioritize prospective validation, algorithmic transparency, and ethical frameworks for real-world deployment in athletic populations.