Among endurance athletes, Achilles tendinitis is one of the most feared injuries, leading to weeks of downtime, persistent biomechanical imbalances, and, in the most severe cases, ruptures. Until now, methods for predicting these injuries have relied primarily on listening to one’s body, interpreting pain, or visually analyzing gait—approaches that are often too late. It is in this context that Stanford has unveiled a groundbreaking advancement. A team of researchers at the university has developed an artificial intelligence model capable of predicting the risk of tendinitis before the first symptoms appear, through the automatic and continuous analysis of biomechanical data from sensors and video recordings1.
AI capable of detecting micro-signals invisible to the human eye
Achilles tendinitis develops gradually, long before any pain is felt. Researchers at Stanford therefore conducted several months of experiments with amateur and professional runners. They recorded their joint angles, stride rate, speed, ground contact, and the cumulative micro-shocks accumulated over the course of many kilometers. The AI model, trained on this vast dataset, can now identify the subtle signs that precede tendinopathy. In particular, it detects unusual variations in stride rate, asymmetries between the left and right legs, abnormal flexion angles, or excessive loads during the stance phase—all factors that would be imperceptible to a coach during a standard observation.
How AI Analyzes Biomechanics in Real Time
To achieve this level of accuracy, the approach relies on a combination of technologies working together. Inertial sensors attached to the legs measure the body’s acceleration and rotation, force platforms analyze ground reaction forces, and high-speed cameras record every movement using a computer vision model. Artificial intelligence then cross-references this information with the athletes’ health data and actual cases of tendinitis observed during the study. It thus generates a dynamic risk score, updated at each session, which indicates whether an athlete is approaching a critical threshold. This system enabled researchers to predict the onset of tendinitis with over 85% accuracy, sometimes several weeks before the first pain, a level of performance previously unattainable2.
Impressive results: early and personalized prediction
The study reveals a major contrast between traditional biomechanical analysis and the predictive capabilities of AI. Whereas a coach observes a few dozen parameters, a deep learning model analyzes several thousand per minute. This difference in scale makes it possible to detect potential micro-injuries before they worsen, paving the way for proactive rather than reactive injury prevention. The model also identified runner profiles that are more vulnerable than others, based on their body type, running technique, or training load, marking a major breakthrough in the personalization of endurance training programs. AI could thus become a cornerstone of sports injury prevention strategies, particularly through wearable devices that directly integrate on-board predictive models3.
Comparison Chart: Traditional Methods vs. Stanford AI
According to Stanford, AI offers a decisive advantage over traditional prevention methods. It anticipates the risk of overload before tissues show the first signs of inflammation.
Here is a summary comparison table:
| Criterion | Traditional methods | AI developed by Stanford |
| Risk detection | Delayed, pain-based | Planning several weeks in advance |
| Accuracy | 50 to 60 percent | > 85 % |
| Parameters analyzed | A few dozen | Several thousand per minute |
| Customization | Limited | Very high |
| Prevention | Responsive | Proactive and predictive |
This approach marks the beginning of a new era in sports medicine, where training decisions are based on objective, real-time data rather than on athletes’ delayed feedback.
Practical applications for athletes and their coaches
This AI opens up new possibilities for endurance training. The concrete benefits include:
- automatic adjustment of training loads based on the risk score,
- detecting mechanical imbalances before they become problematic,
- preventing overloads caused by volume or intensity,
- real-time adjustment of programs based on muscle fatigue,
- recommendations for targeted corrective exercises,
- a reduction in the risk of injury over the long term.
Both amateur and professional athletes could thus extend their athletic careers while improving the quality of their training sessions.
Ethical Issues and Limits: The Line Between Monitoring and Over-Surveillance
The use of biomechanical data raises significant ethical questions. This highly personal information allows researchers to infer not only an individual’s health status but also their physical capabilities and potential performance. Collecting such data requires informed consent, especially in professional sports where hierarchical pressure can influence decisions. Furthermore, the model’s accuracy still depends on the equipment used and the populations studied. Researchers emphasize the need to expand the cohorts, diversify the profiles studied, and ensure that AI is never used as a tool for selection or exclusion. The goal is to enable enhanced prevention, not excessive surveillance.
Conclusion: A New Era in Sports Medicine
The AI developed by Stanford marks a major turning point in the prevention of injuries in endurance sports. By revealing invisible micro-signals and describing an individualized risk profile, it paves the way for safer, more effective training that is better tailored to each athlete’s unique characteristics. Ultimately, this approach could become an essential standard in sports medicine, both for amateurs seeking to improve without injury and for professionals looking to maintain their performance.
Learn more
To learn more about the use of artificial intelligence in the medical field and how it is transforming diagnostic and preventive practices, check out: MedGPT: the free French medical AI that rivals ChatGPT
References
1. Stanford Medicine. (2025). AI-Based Prediction of Achilles Tendinopathy in Endurance Athletes.
https://www.med.stanford.edu
2. Journal of Sports Science and Medicine. (2024). Biomechanical Markers of Achilles Tendon Overload in Runners.
https://www.jssm.org
3. Nature Biomedical Engineering. (2025). Wearable Sensors and AI Models for Predictive Injury Prevention.
https://www.nature.com/natbiomedeng
