What can AI learn from the Sports & Exercise Medicine doctor?
Que peut apprendre l’IA du médecin du sport et de l’exercice ?
Gojanovic Boris
Hôpital de La Tour, Swiss Olympic Medical Center, 1217 Meyrin, Switzerland
Sport & Exercise Medicine Switzerland (SEMS) president
Two years ago, I played a little trick on you when I had Chat-GPT write an editorial on the use of artificial intelligence (AI) in sports medicine. [1] Maybe it would be good to go back to that text before you continue with this special AI issue.
We now live in a world of AI
We must now admit it: we live in a world of AI. There is no ignoring it anymore. [2] Yesterday, I received a video-consultation request from a patient in another country. She not only sent me a few documents and tests results, but she also forwarded a list of medical questions related to further diagnostic testing and therapeutic options. Nowadays, patients come “informed”: she had used a large language model (LLM) to explore her persistent musculoskeletal condition and received with very detailed questions to ask the next clinician she would consult. [3]
In our field of sport & exercise medicine, we usually pride ourselves of being ahead of the curve, embracing innovation and technology. The professional sports world nudges us in that direction, for if we do not travel down that path, athletes will find their own way to the latest gimmicks and therapeutic options promising wonders. However, it seems that AI has many people, including clinicians, sitting back and waiting. Skepticism rightfully prevails, as we have all seen or heard about the hallucinations, the biases, and the imperfections that plague LLMs. On the other hand, we always advocate for better patient/athlete education and preparedness when it comes to discussing their medical issues. AI provides an opportunity to develop a more curious culture whenever we face uncertainty, which what anyone with pain and lack of clear explanations deals with. [4] In the past, we would struggle to hide a little smirk when a patient admitted to having used “Dr Google”. Probably because it was easy to challenge the “knowledge” they had gathered and bring our expert’s point of view to save the day.
We now face a different adversary in Dr Claude, Gemini, Grok, Chat and more. The computation power increases exponentially, the responses are sophisticated and personalized, and the advent of artificial general intelligence (AGI) is looming. [5] Also, our patients use these AI models every day. [4] Which bring this question: what can AI learn from the sports & exercise medicine doctor?
Sport & Exercise Medicine experts challenge AI
The concept of this issue is taking a user-centered approach, with the patient’s journey in mind. A clinical case with fragmented information is fed to a LLM, asking for interpretation and generation of a text which includes interpretation, diagnosis and a full set of recommendation and return to play discussions. This is the part that the patient/athlete would potentially do.
Next, we asked the experts to review the generated text and discuss the gaps and errors to come to a proper clinical assessment. This is the work you would do at the consultation when the patient comes with her AI-based knowledge.
The articles generated by AI are by design imperfect. First, we can except the inherent errors of the models. Second, to reach the best possible result, AI should be used in multiple iterations to optimize the result through repeat-prompting with specificities warranted by each clinical case. We assumed that the patient would not do this, and that this is the skill of the clinician.
A word of encouragement and… of caution
I encourage you to go through all these cases and to challenge your preferred AI models with your own cases, to get a better grip on what it can and cannot do. I will add a word of caution: the articles were generated over the course of 1 month. During that short time span, the main AI companies released their latest models, with significant improvements. We have never seen such a fast and exponential speed of development of new technologies. Imagine waiting for sunrise in a beautiful spot in nature. You can literally see the sun rise and the light gradually bathe your surroundings. This is what is happening with AI today, it rises quicker than we think and its capacities grow very fast, scarily fast.
Last but not least, I would like to thank all the expert clinicians and Carina Brunner of Swiss Sport Integrity who agreed to play along for this hybrid approach to articles in your SEMS journal. Enjoy the ride.
Correspondence
Boris Gojanovic
President Sport & Exercise Medicine Switzerland (SEMS)
Hôpital de La Tour, médecine du sport
av. J.-D. Maillard 3
1217 Meyrin (GE), Switzerland
Email: boris.gojanovic@latour.ch
References
- Gojanovic B. Bridging the Gap: Artificial Intelligence in Sports Medicine and Musculoskeletal Rehabilitation. Sport & Exercise Medicine Switzerland Journal. 2023;71(4):4-6.
- Cheng K et al. Artificial Intelligence in Sports Medicine: Could GPT-4 Make Human Doctors Obsolete? Ann Biomed Eng. 2023 Aug;51(8):1658-1662.
- Morgan DJ, Rodman A et Goodman KE. How Physicians Can Prepare for Generative AI. JAMA. 2025;185;(12):1407-1408.
- Ramkumar PN et al. Sports Medicine and Artificial Intelligence: A Primer. American Journal of Sports Medicine. 2022 Mar;50(4):1166-1174.
- Grace K et al. Thousands of AI Authors on the Future of AI. Journal of Artificial Intelligence Research. 2025;84:9.
Comments are closed.