IPM Take
Sudden cardiac death in young athletes is rare, but when it happens, it becomes a public tragedy and a policy failure in real time. The promise of explainable AI is not simply that it may detect hidden risk. It is that it could force a harder debate about what modern prevention should look like: who gets screened, how often, using which data, with what safeguards, and who carries responsibility when the algorithm is wrong. Athlete cardiac safety cannot become a luxury service for elite clubs while schools, community sport and lower-resource systems are left with guesswork.
Executive Summary
A new systematic review published in npj Digital Medicine examined sports-related sudden cardiac death and sudden cardiac arrest in adolescents and young adults, alongside explainable AI methods and cardiac electrophysiological models. The review identified 9,574 studies and included 84, covering epidemiology, AI approaches and cardiac modelling. Sports-related sudden cardiac death incidence ranged from 0.1 to 0.6 per 100,000 participants per year. Gradient-weighted Class Activation Mapping was the most common explainable AI technique, while cardiac electrophysiological models mainly focused on cellular and tissue-level mechanisms. The authors argue that standardised definitions and better integration of epidemiological risk factors, explainable AI and cardiac modelling are needed to advance athlete-specific risk stratification.
Why it matters
- Policymakers: Athlete screening cannot be modernised through technology alone. National frameworks will need rules on validation, consent, liability, exclusion from sport and equitable access.
- Clinicians: Explainable AI could support interpretation of ECG, imaging and arrhythmia risk signals, but it must be clinically validated and understandable enough to support shared decision-making.
- Data / AI leaders: Black-box prediction is not enough in sports cardiology. Any tool used to influence eligibility, training or return-to-play decisions will need transparency, auditability and bias testing.
- Industry / innovation partners: There is a clear innovation opportunity for validated AI-enabled cardiac screening and digital heart modelling, but prospective evidence and real-world implementation data will decide credibility.
- Patients / athletes: False negatives can be fatal. False positives can end careers, restrict participation and cause psychological harm. Prevention policy must protect lives without turning risk prediction into automatic exclusion.
Athlete cardiac screening is entering a new phase.
For years, the policy debate has circled around familiar questions: ECG or no ECG, elite athletes or all athletes, one-off screening or repeated monitoring, national programme or club-level responsibility. Now AI is entering the arena, and the stakes are higher.
A systematic review published in npj Digital Medicine brings together three fields that have often moved separately: the epidemiology of sports-related sudden cardiac death and sudden cardiac arrest, explainable AI for life-threatening arrhythmias, and cardiac electrophysiological modelling.
The review found that sports-related sudden cardiac death in adolescents and young adults remains rare, with incidence estimates ranging from 0.1 to 0.6 per 100,000 participants per year. But rarity is not reassurance. These events often occur without warning, affect otherwise healthy young people, and trigger intense public scrutiny because they unfold in schools, clubs, stadiums and broadcast sport.
The recent collapse of Christian Eriksen during Denmark’s friendly against Ukraine, after his implantable cardioverter-defibrillator delivered a shock, brought the issue back into public view. His case is not the same as undetected risk in an unscreened athlete. But it exposed the same uncomfortable policy problem: cardiac risk in sport is not abstract. It is lived in real time by athletes, families, clinicians and sporting institutions.
The review suggests that explainable AI could become part of the answer. Unlike black-box systems, explainable AI is designed to show which signals contributed to a prediction. That matters in cardiology, where a risk score can influence whether an athlete trains, competes, undergoes further testing or is advised to stop.
The review found that one of the most widely used explainable AI approaches helps show why an algorithm reached a particular conclusion by highlighting the signals or patterns that influenced its decision. Alongside this, researchers are using digital models of the heart that can simulate how electrical signals move through heart tissue. These virtual heart models can help scientists better understand how dangerous abnormal heart rhythms develop and why some athletes may be at higher risk than others.
The policy promise is powerful: combine population risk factors, clinical data, AI interpretation and digital heart modelling to move from blunt screening to personalised risk stratification.
But this is where the politics begins.
A screening tool is never just a tool. It creates consequences. Who gets access to advanced screening? Who pays for it? What happens when an algorithm flags a young athlete as high risk? Can that athlete appeal? Who explains uncertainty to the family? Who is liable if a tool misses a risk signal? And how do systems avoid embedding bias if training datasets overrepresent certain populations, sports, ethnic groups or elite athlete cohorts?
The review also highlights a basic but serious problem: definitions of sports-related sudden cardiac arrest and sudden cardiac death vary across studies. Without standardisation, evidence cannot translate cleanly into policy, reimbursement or guidelines. Fragmented definitions produce fragmented systems.
For IPM, the message is clear. AI-enabled athlete screening should not be sold as a miracle layer added on top of weak infrastructure. It needs a policy pathway: agreed definitions, prospective validation, transparent models, data governance, referral capacity, psychological support and equity safeguards.
Otherwise, the future of athlete cardiac safety will divide into two systems: one where elite athletes are monitored with cutting-edge digital tools, and another where everyone else depends on luck.
