How artificial intelligence can revolutionize sport: coaches touch algorithms to improve performance, prevent injuries

This data revolution will mean that some overuse injuries could be significantly reduced in the future, says Stephen Smith, CEO and founder of Kitman Labs, a data company that operates in several professional sports leagues with offices in Silicon Valley and Dublin. “There are athletes who treat their bodies as a job and have started using data and information to better manage themselves,” he says. “We will see many more athletes who will play much longer and play at the highest level much longer.

While offering prospects for preserving players’ health, this new frontier of artificial intelligence and sport also raises difficult questions about who will possess this valuable information - individual athletes or team managers and coaches who benefit from the data. Concerns about privacy also arise.

A baseball application called Mustard is among those that already use computer vision. Videos recorded and sent by users are compared to a database of professional thrower movements, prompting the app to suggest prescribed exercises aimed at helping to throw more efficiently. The Mustard, which comes in a free download, is designed to help ambitious players improve their performance, as well as avoid the kind of repetitive movements that can cause long-term pain and injury, says CEO and co-founder Rocky Collis.

Computer vision is also entering applications for other sports, such as golf, and promises to be relevant for amateurs and professionals in the future. Algorithms using a form of artificial intelligence known as machine learning, which collects statistics from sensors and can analyze changes in body position or movement that could indicate fatigue, weakness or potential injury, are now in widespread use. The Liverpool football club in Great Britain says that it has reduced the number of injuries of its players by a third compared to last season, after it adopted a data analysis program based on the artificial intelligence of the company Zone7. The information is used to customize training recipes and suggest optimal rest times.

Football is among the biggest adopters of data analytics driven by artificial intelligence while teams are looking for any advantage in global sports. But some individual sports are also beginning to use these technologies. At the 2022 Winter Olympics in Beijing, ten American skaters used a system called 4D Motion, developed by New Jersey’s 4D Motion Sports, to help track the fatigue that can result from too many leaps in practice, says Lindsay Slater. manager of sports sciences for figure skating in the United States and assistant professor of physical therapy at the University of Illinois at Chicago. The skaters attached a small device to their hips, and then reviewed the movement data with their trainer as they practiced.

“We’ve actually brought the algorithm to the point where we can really define the take-off and landing of a jump, and we can estimate that the stresses on the hook and torso are quite high,” says Dr. Slater. “During the day, we found that athletes have reduced angular velocity, reduced jump height, cheating multiple jumps, where these chronic injuries and injuries from overuse usually occur.” She says American figure skating is evaluating the 4D system in a pilot project before expanding its use to more of its athletes.

Algorithms still have to overcome many obstacles in predicting injury risk. First of all, it is difficult to collect long-term data from athletes who jump from team to team every few years. Also, the data collected by the sensors may vary slightly depending on the device manufacturer, while the visual data has the advantage of being collected remotely, without worrying that the sensor could break down, say analysts.

Psychological and emotional factors that affect performance cannot be easily measured: stress during contract negotiations, quarrel with spouse, bad food the night before. And the only way to truly test algorithms is to see if a player who is marked as an AI program is really injured in the game - a test that would violate ethical rules, says Devin Pleuler, director of analytics in Toronto. FC, one of the 28 teams in the Major League Soccer.

“I think there could be a future where these things can be trusted and trusted,” he said. Players. “But I think there are significant issues with sample size and ethical issues that we need to overcome before we can really reach that kind of threshold.

Both data privacy issues and the question of whether individual athletes should be compensated when teams gather their information to power artificial intelligence algorithms are also challenging.

The U.S. currently has no regulations prohibiting companies from collecting and using player training data, says Adam Solander, a Washington attorney who represents several major sports teams and data analysis firms. He notes that the White House is developing recommendations on rules governing artificial intelligence and the use of private data.

Those regulations will have to strike a balance so that potentially important technologies can help people, while respecting the privacy rights of individuals, Solander says.

For now, one sports data firm that has adopted computer vision is using it not to predict injuries, but to predict the next superstar. Paris-based SkillCorner collects television videos from 45 football leagues around the world and runs them through an algorithm that tracks the location and speed of individual players, says Paul Nilsson, the company’s CEO.

The company’s 65 clients now use the data to reconnoiter potential recruits, but Mr. Nilsson expects that in the near future, the company’s video games could be used in an effort to identify injuries before they occur. However, he doubts that the AI ​​algorithm will ever replace the human trainer on the side.

“During the game, you’re there and you can smell it, feel it, almost touch it,” he says. “For these decision-makers, I think it is even less likely that they will really listen to the insight that comes from the source of artificial intelligence.