Cross-cultural validation of the Chinese version of the lean management scale for nursing services in hospitals.

Machine learning and artificial intelligence have reshaped how risk is quantified, performance modelled, and care delivered across sports medicine over the past decade. Systematic reviews confirm that machine learning algorithms can identify athletes at elevated injury risk by analysing training load metrics, biomechanical patterns, and physiological markers (, ). Wearable biosensors generate continuous streams of physiological, biomechanical, and biochemical data that adaptive ML platforms convert into recovery readiness scores, fatigue indicators, and individualised load recommendations (, ). In team sports, integrating GPS telemetry, heart rate variability, and accelerometry into adaptive algorithms has enabled coaches and medical staff to make evidence-informed training decisions with a level of precision unavailable to previous generations of practitioners (, ). AI-powered diagnostic systems, computer vision movement analysis, and natural language processing of electronic health records further extended the clinical reach of these technologies (, ). Topol and colleagues have proposed the concept of high-performance precision medicine as an organising principle for next-generation athlete health management, based on the convergence of multimodal data and adaptive algorithms (). Concurrently, the operational challenges of implementing AI within multidisciplinary sports science and medicine teams became an active area of investigation (), as practitioners recognised that strong performance on benchmark datasets (i.e., standardised evaluation datasets used to assess AI model performance, such as medical licensing examination question sets) does not guarantee clinical utility under conditions of variable data quality, institutional constraint, and absent human oversight (). These concerns, raised by sports medicine physicians, physiotherapists, strength and conditioning coaches, and sporting organisation administrators, crystallised around two issues: the risk of deploying tools that perform well on benchmark datasets but fail on domain-specific tasks (e.g., individualised return-to-sport decisions after injury or tailored exercise prescription for athletes with chronic conditions), and the absence of governance frameworks that align AI-augmented practice with professional ethics and patient safety obligations. These concerns are grounded in emerging evidence, including documented rates of LLM hallucination and associated risks to clinical reliability and research integrity (). The public release of ChatGPT in November 2022 introduced a qualitatively different category of AI to sports medicine.