Artificial intelligence (AI) is rapidly reshaping medicine, from diagnosing disease to accelerating drug discovery. Its influence is also reaching the world of sport.
In a recent editorial , we looked at how AI could transform how researchers, doctors and sporting organisations detect, monitor and manage concussion, which is one of the most challenging health issues facing contact sports today.
Concussion is a form of brain injury caused by a sudden bump, blow, or violent shake to the head. It is difficult to diagnose as it rarely looks the same twice. Some athletes feel sick and dizzy; others do not. Some lose consciousness, yet around 90% do not .
Many of the sports we play and watch carry a significant risk of head impact and brain injury. Such injuries might occur from a single heavy impact: a mistimed tackle, a punch, a fall. Others add up over many years of frequent knocks, leaving athletes at an increased risk of Parkinson's disease, Alzheimer's or other forms of dementia.
Our own research has shown how even repeated impacts that don't cause concussion in rugby , football and boxing can quietly damage the brain's blood supply and function over time.
How does this help athletes?
The more precisely we can pinpoint which parts of the brain have been affected by an impact, the better we can tailor an athlete's recovery. Rather than applying a one size fits all checklist for returning to play, doctors and sports staff can instead use data-backed insights to personalise rehabilitation plans. In doing so, they should track not just the player's physical recovery, but also their psychological readiness to return to action.
AI could also help with one of the thorniest issues in concussion management: the pressure placed on athletes to return too soon by clubs, coaches and themselves. An independent AI model, drawing on a range of data from brain scans and blood tests to surveys of an athlete's mood, could provide medical staff with a solid foundation of objective evidence to resist such pressures.
We are already exploring these approaches, particularly in objective blood and saliva testing, in collaborative research with the charity Head for Change. The organisation supports former athletes living with neurodegenerative conditions to better understand long-term brain health in contact sport.
AI can also turn data from wearable sensors in helmets and gumshields, for example, into maps of the brain's injuries . This matters because every athlete is different. Factors like neck strength, fatigue and previous injury history mean that a single hit inflicts varying damage in different people.
Like any tool, AI is not without its risks. It can sometimes present false claims as definite facts. If scientists rely on these summaries, they may build studies on flawed foundations. There is also the danger of false reassurance, where a tool incorrectly labels an injury as low risk for concussion and an athlete is returned to play too soon.
Ethical hurdles involve both old and new data. AI is only as reliable as its training material, resulting in observed gender and racial biases . If models use historical data from, say, solely male professionals, they may fail women, children or amateur players. We also face questions regarding athletes' medical data, and whether it is owned by the player, club, or even the insurer.
Perhaps the subtler danger is one for scientific culture. As pressure mounts on academics to publish ever more material, there is a risk that leaning too heavily on AI as a shortcut will dampen genuine curiosity and creativity. This would produce, as one recent paper put it, a situation in which we "produce more but understand less".
Where does this leave us?
AI is not going to replace doctors, physiotherapists or the careful human judgement required to manage a concussed athlete. It should not be used as a shortcut to simply return athletes to the pitch sooner either.
But used wisely, it could help health professionals make better decisions: spotting injuries earlier, tracking recovery more precisely and protecting long-term brain health in ways that were not previously possible.
The challenge now is to build the appropriate guardrails like open, transparent AI systems that can be interrogated and held accountable, trained on diverse and representative data. Done right, this technology could become one of our most powerful tools to protect the athletes competing in our favourite sports.
Damian Bailey is supported by a Royal Society Wolfson Research Fellowship (Grant No. WM170007). He is Editor-in-Chief of Experimental Physiology and outgoing Chair of the Life Sciences Working Group and outgoing member of the Human Spaceflight and Exploration Science Advisory Committee to ESA. Damian is also a current member of the Space Exploration Advisory Committees to the UK and Swedish National Space Agencies, and a consultant to the brain health charity Head for Change.
Danny William Walmsley does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.
