Getafe CF has one of the lowest operating budgets in Spain’s top soccer circuit, La Liga. That puts a premium on the club keeping its starting players healthy. Depth is hard to afford.
The team used to endure about three notable injuries per month, but something curious happened last year: that rate fell to one per month. The club had a 65-percent reduction in days missed due to injury. Some 80 percent of the injuries that still occurred also came with identifiable warnings, indicating that there is room for improvement and that benching an important player can be a difficult decision.
Those metrics—and the underlying injury prediction system—were produced by Zone7, a new artificial intelligence platform that inputs data from wearables devices, fitness assessments, and medical profiles, then uses that information to determine which players might be at risk of injury.
“We felt that, counterintuitive to a lot of the industry’s mainstream thinking, injury prediction is something that you can actually predict,” said Zone7 cofounder and CEO Tal Brown. “But you need to be very focused on what you’re trying to predict. It’s not a case of ‘at 4 o’clock a month from now, you’re going to tear a hamstring.’ You need to apply some product thinking and make this usable by a coach.”
Brown and his cofounder, CTO Eyal Eliakim, are both veterans of the Israel Defense Force’s intelligence corps, where they were Big Data experts. They later applied that knowledge while working at Salesforce before founding Zone7. The new injury analytics company works with five European soccer franchises, two MLS clubs, six NCAA teams (including soccer, lacrosse, and field hockey), a national Olympic team, and one MLB franchise.
Brown and Eliakim have their own sporting backgrounds. Brown is a black belt in aikido and was a martial arts instructor for 10 years. Eliakim played high level juniors tennis.
Jordi Cruyff played at Barcelona and Manchester United before working as a sports director and coach. (He currently manages Chongqing Dangdai Lifan F.C. in the Chinese Super League.) Cruyff worked with Zone7 in a previous job. The system had generated warnings for five of seven key injuries that club suffered, which he said was clearly “not a coincidence.” He has since collaborated with Zone7 as an adviser and investor.
“I’m not a big specialist in this kind of technology,” Cruyff said, “but I did see with my own eyes the results that it had.”
During that first pilot test, the club’s manager did not heed the Zone7 warnings. After seeing the results, however, Cruyff said that he ensured action was taken upon receiving future alerts. The system provides red, yellow, and green indicators for each player’s daily risk level.
“We would down-load trainings for that player,” he said. “We tried to make sure that we would not push him so hard that he would get injured. We would down-load him in the trainings to have him fresh and to minimize risk for injuries.”
Brown said prior injuries are the most critical indicator of future ailments, but data culled from GPS devices such as Catapult or STATSports as well as heart rate monitors are among the other important sources of information. According to Brown, most workout tracking systems categorize load in six zones. The name Zone7 was derived from this platform’s attempt to go beyond the norm. The platform works with a range of external devices and can be adapted for all sports, although the focus now is on outdoor field sports.
“One of the benefits of being unobtrusive as a concept for the product is that we’re not asking them to do anything different,” Brown said. “We’re not slapping on any hardware. We’re not asking them so step out of the training gym and do some kind of special test on a special hardware. But they do need to follow the advice.”
The software will grow more sophisticated as more athletes are entered into the database. Roughly one million training sessions and injuries having already been recorded and analyzed. Zone7’s website currently touts a 95-percent accuracy rate and 75-percent reduction in injuries.
“If you look at a sport that has a weekly cadence of games like soccer, an interesting question that you want to answer is, ‘Who’s predicted to be injured next week?’” Brown said. “A lot of tools today will try to give an indication of ‘Yeah, this guy’s running too much’ and there are all these parameters you can look at in terms of speed and distance and heart rate. But what we try to do is compare every athlete’s current state to a bank, or database, of previous injuries that actually occurred to other people.”
While Zone7’s algorithms would seem most pertinent to non-contact injuries, Brown said a state of advanced fatigue can also lead to contact injuries.
“A fatigued player will react differently to contact than a non-fatigued player,” he said. “What we are seeing is that—intuitively, what we thought would be mostly relevant to non-contact injuries—we’re actually seeing a lot of contact injuries being predicted as well.”
At the end of the day, Brown said, what’s important is distilling the information clearly and succinctly. As such, the thrust of Zone7 is not on fancy visuals or an elaborate user interface.
Added Brown, “It’s about, ‘Can the guy trust what you’re telling them?’”