EDITORIAL

published: 25 April 2022

doi: 10.3389/fspor.2022.886730

Editorial: Using Artificial Intelligence

to Enhance Sport Performance
Dan B. Dwyer 1*, Matthias Kempe 2 and Arno Knobbe 2,3
1
School of Exercise and Nutrition Sciences, Deakin University, Geelong, VIC, Australia, 2 Department for Human Movement
Sciences, University of Groningen, Groningen, Netherlands, 3 Leiden Institute of Advanced Computer Science, Universiteit
Leiden, Leiden, Netherlands

Keywords: sport, artificial intelligence, performance, machine learning, decision making

Editorial on the Research Topic

Using Artificial Intelligence to Enhance Sport Performance

The purpose of this special topic is to provide an opportunity for researchers and practitioners to
showcase the most recent advances in machine learning and artificial intelligence in sport. The
relatively recent expansion of technology in sport has enabled an explosion in the amount of
data collected and also the range and variety of attributes that are measured and recorded. This
tsunami of data has created both a burden and an opportunity to answer some critically important
questions that are faced by coaches and athletes. How can I train more effectively, how can I be more
competitive and how can I avoid injury? In many cases the applicability of traditional statistical
techniques has been exhausted and machine learning methods have been applied, adapted and
developed, to analyse sport data.
Edited and reviewed by:
In 1995, Lapham and Bartlett described artificial intelligence (AI) as holding the potential to
Kamiar Aminian,
Swiss Federal Institute of Technology support and improve decision-making in sports and speed up the analytic process to free up time
Lausanne, Switzerland and resources for experts. The use of Machine Learning (ML) has been advocated since then to
*Correspondence:
build such decision support systems (Robertson, 2020). ML can be applied in sport to create many
Dan B. Dwyer benefits including; to automate or semi-automate the collection of data, to (pre-)process data into
[email protected] meaningful information, to understand what information is important with respect to health and
performance, and finally, to help coaches and athletes make complex decisions. Elite coaches and
Specialty section: athletes frequently use their experience, knowledge and intuition to make successful decisions.
This article was submitted to However, some critical decisions are very difficult because the number of factors to consider and
Sports Science, Technology and their interactions are too complex. Where the right kind of data are available, Machine Learning
Engineering, methods can be used to create models that can support complex decision-making.
a section of the journal
Many of the articles in this Research Topic represent the adaptation of existing, and the
Frontiers in Sports and Active Living
development of novel ML methods, that solve real and significant problems in sport and confer
Received: 28 February 2022 many of the benefits mentioned above. In terms of improving the gathering and processing of data,
Accepted: 17 March 2022
Hosp et al. created a deep learning tool to automatically classify the level perceptual performance
Published: 25 April 2022
expertise of football (soccer) goalkeepers, while van Dijk et al. described how their advancements
Citation:
on the processing of data from inertial measurement units, produces more accurate information
Dwyer DB, Kempe M and Knobbe A
(2022) Editorial: Using Artificial
about body position in wheelchair sports. Furthermore, Schmid et al. created a tool to automatically
Intelligence to Enhance Sport identify tactical pattern-based player trajectories in American Football and used these automated
Performance. annotations to create a simulation for defensive schemes and performance.
Front. Sports Act. Living 4:886730. The development of new measures to evaluate individual and team performance was
doi: 10.3389/fspor.2022.886730 the main point of the majority of the studies in this Research Topic. Anzer and Bauer

Frontiers in Sports and Active Living | www.frontiersin.org 1 April 2022 | Volume 4 | Article 886730
Dwyer et al. Editorial: Artificial Intelligence in Sport

presented an improved method for calculating expected goals decision making. The application of ML in sport has been
(i.e., scoring probability) in football (soccer) which is a well justified based on what it can offer sport. However, we may
established measure of individual and team performance. The soon reach the conclusion that sport offers something uniquely
concept of “zones of control” is an important concept in the valuable to Machine Learning. Sport can provide high quality,
analysis and understanding of tactical performance in invasion rich data sets about many different human behaviors (e.g.,
sports such as football. Martens et al. described a more advanced movement, actions, performance, interactions, health, etc.) from
method to calculate the control of space which has the potential a variety of different contexts. The behavior and performance
to improve the way offensive performance is analyzed. This idea of athletes are constrained by clearly-defined rules and trained
was closely linked to Dick et al. who explored ball possession experts can establish ground truths to support supervised
phases in football in order to better understand how player Machine Learning. Many of these characteristics do not apply in
behavior determines the outcome of the phase. They presented a business context where much of the development of Machine
a new set of performance indicators that can be used to coach Learning occurs.
and assess player performance. Our hope is that the development of Machine Learning in
The last group of papers created models to predict athletes’ sport continues, by educating students and practitioners about
performance, including Olthof et al. who explored the complex its benefits and applicability to sport. Indeed, each field (sport
relationship between biomechanical load and performance in and data science) is sufficiently complex, and it may be wise to
basketball players before and during games. They revealed that move ahead by creating partnerships between Sport Scientists
game load is a good predictor of technical performance and and Data Scientists. We need the leaders in this field to continue
training load 2 days before gameday is a good predictor of to drive innovation but also to provide guidance toward the most
game load. Finally, Kholkine et al. developed a method to help beneficial solutions and away from ethically risky applications.
coaches understand who the main adversaries would be in 1-day Finally, we look forward to seeing the machine learning solutions
road cycling races. Their prediction model can also help identify that have been developed, deployed in the field to work for
the important factors for winning a specific race and adapt the us by collecting data, conducting preliminary analyses and
strategy of the team if certain riders display patterns of possibly making recommendations.
winning a future race.
Every article in this Research Topic represents a valuable AUTHOR CONTRIBUTIONS
advancement in ML in sport. Collectively, the findings of the
articles improve how we process raw data, clarify which variables DD wrote the first draft, then MK and AK edited this draft to
represent the most important indicators of performance and create the final version. All authors designed the structure and
provide prediction models that can be used to support complex content of the editorial.

REFERENCES Publisher’s Note: All claims expressed in this article are solely those of the authors
and do not necessarily represent those of their affiliated organizations, or those of
Lapham, A. C., and Bartlett, R. M. (1995). The use of artificial intelligence in the the publisher, the editors and the reviewers. Any product that may be evaluated in
analysis of sports performance: a review of applications in human gait analysis this article, or claim that may be made by its manufacturer, is not guaranteed or
and future directions for sports biomechanics. J. Sports Sci. 13, 229–237.
endorsed by the publisher.
doi: 10.1080/02640419508732232
Robertson, S. (2020). Man and machine: Adaptive tools for the
Copyright © 2022 Dwyer, Kempe and Knobbe. This is an open-access article
contemporary performance analyst. J. Sports Sci. 13, 2118–2126.
distributed under the terms of the Creative Commons Attribution License (CC BY).
doi: 10.1080/02640414.2020.1774143
The use, distribution or reproduction in other forums is permitted, provided the
original author(s) and the copyright owner(s) are credited and that the original
Conflict of Interest: The authors declare that the research was conducted in the publication in this journal is cited, in accordance with accepted academic practice.
absence of any commercial or financial relationships that could be construed as a No use, distribution or reproduction is permitted which does not comply with these
potential conflict of interest. terms.

Frontiers in Sports and Active Living | www.frontiersin.org 2 April 2022 | Volume 4 | Article 886730