When the University of Toronto Tanenbaum Institute for Science in Sport (TISS) was launched in May 2022, director Ira Jacobs said that one of the things he is most looking forward to with regards to what TISS will enable, is the opportunity for sport scientists to communicate with each other, while giving those who are in the high-performance world – athletes, coaches, etc. – an opportunity to receive cutting edge and applicable new knowledge.
On September 13, Jacobs, who is a professor and former dean at the U of T Faculty of Kinesiology and Physical Education (KPE), hosted the institute’s first speaker series with a panel discussion on the potential of artificial intelligence (AI) within sport science, sport medicine and sport analytics research.
The panel featured Joseph (Joe) Baker, a professor at KPE and Tanenbaum chair in sport sciences, data modelling and sport analytics, Daniel (Danny) Whelan, an associate professor at the Temerty Faculty of Medicine, who holds the Tanenbaum professorship in orthopaedic sports medicine, and Meghan Chayka, co-founder of Stathletes and data scientist in residence at the Rotman School of Management.
Taking in the discussion, just like Jacobs had envisioned, was a cross section of the high-performance world, including scholars and surgeons, athletes and coaches, policy makers and reps from professional sport leagues and sport associations.
We caught up with Baker after the event to chat about the potential applications of AI in sport science, how it can be used to optimize athlete performance, and whether the benefits of integrating AI into sport science outweigh the challenges.
What are some potential applications of AI in sport science?
Given how dependent sport scientists and practitioners are on information for effective decision-making, training design and so on, the applications are nearly limitless. We’ve seen the influence of AI on everything from advanced data modelling for talent identification and player forecasting to designing better training environments and creating superior ways to monitor athlete stress, learning and performance.
How is AI transforming the way sports-related injuries are diagnosed? What role can AI play in enhancing the treatment and rehabilitation process for athletes?
We have seen a lot of attention in many sports on injury diagnosis, for example for concussion, and injury prediction, especially at the elite and professional levels. For many professional teams, a model that could predict an athlete’s injury risk is seen as the ‘holy grail’ of data analytics.
How can AI be used to optimize athlete performance?
An athlete’s ability to perform in an optimal way at a specific time reflects a complex interaction of physiological, biomechanical and psychological systems, among others. Until recently, researchers haven’t had the statistical and computing tools to be able to explore these complicated relationships. AI has the potential to help us understand the interaction of these systems over time and across development so that we can build stronger models of athlete skill acquisition and performance.
What challenges do you foresee with the integration of AI into sports research? Do they outweigh the potential benefits?
The greatest challenge to the use of AI in sport settings is access to data. Most AI-based approaches, for example, machine learning, require very large datasets to ‘learn’ the most effective patterns of data to look for. Unfortunately, in most sport settings, especially those involving more elite populations, the samples are quite small, for example, very few become professional athletes or Olympians. In addition, sport performance is continually evolving. In sport contexts, it’s not clear whether the benefits of AI will be able to offset these challenges.