About this Research Topic
Success in sport often requires making good decisions about complex issues based on constantly changing information. Practitioners must recruit players based on forecasts about their future performance potential, or design training programs and manage training to balance the need for improvements in performance with the need to keep injury risk low. There are also complex decisions to be made about how to compete, with respect to the technical and tactical aspects of play. While coaches are very capable of making good decisions based on their experience and opinions, these decisions might be improved if they were supported by evidence. Especially given the increasing amount of new data sources that could be used to evaluate athletes, such complex decisions should be made on the available evidence and on the experience that lies in the historical data.
Furthermore, the decisions that can be supported are becoming much more fine-grained, with coaches and support staff having to make many micro-decisions on a daily basis about a large variety of issues, such as training load, tactical plans, scouting, etc. Optimising performance in sports is becoming a matter of getting as many of these micro-decisions right as possible, based on substantial and detailed data, as well as well-founded Machine Learning methods. Part of this development is a shift towards personalised modelling of athletic performance. Sports science has already established different sports disciplines requiring different body-types and appropriate training regimes (e.g. sprint vs. endurance training). However, personalisation allows a much more fine-grained optimisation, exploiting the historical data of individuals, to understand their specific strengths and weaknesses, for example where injuries are concerned.
On the technical side, sports data offers an interesting test bed for data mining researchers. The universe of sports data is not discrete, and the elements of human individuality and human error have more influence, increasing the challenge for researchers. Yet at the same time, many sports are constrained by rules and the place where the action takes place (track, pitch, ice rink, etc.), offering natural experiments and repeatability that other real-life settings lack. Therefore, sports data is really attractive for computer science research to create new methods and apply them to a real-world problem.
In recent years, Machine Learning methods often provide a better opportunity to answer complex questions and discover unknown but important relationships within datasets. While Machine Learning methods have been applied to sports data for many years, there is an ongoing need for the development and refinement of methods and a greater understanding of the implementation of artificial intelligence in the sport context.
Artificial intelligence and machine learning methods can play a crucial role in sports practice in the future. They are beneficial for the entire process of decision making by speeding up and automating the collection of information, pointing out patterns in the information, combining and simplifying large and complex data sets, enabling understandable visualisation of information, and providing automated decision support.
The topics of interest in this Research Topic include;
• Solutions for data analysis problems associated with complex datasets in sport
• Novel approaches for automated information extraction from streaming data (like event detection)
• Overviews on the use of AI in sports practice
• Comparing and contrasting techniques to support decision making in sport
• Exploration of ethical issues in the use of AI in sport
• Personalised models for load monitoring and injury prevention
Keywords: Data science, data mining, machine learning, modelling, predictive analytics, pattern recognition
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.