With just days to go before the start of the Milano Cortina 2026 Olympic Winter Games, artificial intelligence is consolidating its role as a strategic tool in managing physical risk in elite sport. Beyond competitive performance, federations, leagues and clubs are integrating predictive models, digital twins and advanced biomechanical systems with a shared objective: to anticipate injuries before they occur and reduce exposure to structural damage in increasingly congested calendars.
At USA Hockey, the national governing body with 1.2 million members, the starting point has been the digitisation and analysis of injury claims data linked to its captive insurance programme. Under the financial leadership of Kelly Mahncke, the organisation has begun collecting detailed information on when in a game an injury occurs, where on the ice it happens, which body part is affected and whether a penalty was involved. “If you prevent even one injury, it means the world to that person and their family,” Mahncke said. The aim is to detect patterns by age group, phase of play or type of impact and feed those insights back to coaches, officials and equipment manufacturers.
From historical data to injury prediction
In the National Football League -NFL-, the approach centres on the Digital Athlete system, a platform that combines video, training and game data with cloud technology from Amazon Web Services -AWS- to run millions of game simulations. All 32 clubs have access to dashboards providing daily training load metrics, injury risk indicators and league-wide benchmarks. The model is used not only to individualise recovery and prevention programmes but also to assess rule changes, such as the dynamic kickoff, after simulating the equivalent of 10,000 seasons to evaluate potential reductions in injury rates.
Within the National Basketball Association -NBA-, a rise in Achilles tendon ruptures during the 2023–24 season has accelerated the development of a centralised AI-driven system. The initiative aggregates data from team medical staff, video footage, wearables and performance records into a league-wide repository designed to flag biomechanical warning signs. More than 500 players have already undergone biomechanical assessments, with additional rounds of testing scheduled. Commissioner Adam Silver has indicated that the objective is to expand predictive capabilities to other high-risk injuries, including anterior cruciate ligament tears, hamstring strains and overuse conditions.
Digital twins and precision sports medicine
At club level, FC Barcelona, through its innovation ecosystem, has advanced the development of individual digital twins for its athletes. Built on the centralised biomedical platform of Genomcore, the system integrates genetic, metabolomic and proteomic data alongside internal and external load, sleep patterns, nutrition and psycho-emotional context. The intelligence layer developed by Made of Genes transforms this multimodal dataset into early warning signals — referred to as “orange flags” — enabling intervention before a “red flag” or manifest injury occurs. The model seeks to identify subtle deviations in markers such as lactate, cortisol, hormonal fluctuations or accumulated load, combining variables that in isolation would not be conclusive.
Digital twin applications extend beyond football. In France, institutions working with the national cycling federation have developed the Margaria-Morton model to prevent fatigue by simulating the body’s three primary energy systems: phosphocreatine, anaerobic metabolism producing lactate and the aerobic system. Using laboratory tests — including maximal three-minute efforts, VO₂max measurements and instrumented power sensors — researchers created personalised avatars capable of predicting metabolite evolution with an error margin of approximately 15 to 16 per cent. In real-time applications, systems based on the TRIMP (Training Impulse) metric compare expected and actual load to adjust sessions dynamically and prevent overtraining.
From performance to holistic athlete wellbeing
In Formula 1, Scuderia Ferrari HP has integrated physiological monitoring devices from WHOOP to analyse sleep, stress and recovery in both drivers and team members. Medical staff work alongside performance science specialists to interpret the data within an engineering-driven approach applied to the human factor. According to Lorenzo Giorgetti, Chief Racing Revenue Officer of the team, the partnership extends Ferrari’s data-driven philosophy “beyond the car, to the human component.” In an environment characterised by high cardiovascular demand, heat exposure and travel fatigue, continuous monitoring aims to optimise physical availability throughout the season.
From ice hockey to basketball, football, cycling and motorsport, the trend converges on a shared paradigm: integrating large-scale clinical and performance datasets into artificial intelligence models capable of identifying non-obvious relationships between variables and issuing early alerts. The shift from reactive medical treatment to predictive risk management is reshaping training design, reducing uncertainty and positioning injury prevention as a structural pillar in the governance of elite sport.
