The NFL hopes big data tools can help reduce concussions, ligament tears and other injuries in every professional football game. Currently, the number of injuries per game is stable at an average of six or seven. The Wall Street Journal reported this week that NFL engineers are working with Amazon Web Services to apply machine learning and artificial intelligence tools to player data in the hope of finding games that typically lead to injuries.
“Ultimately, we will be able to determine the risk of injury and be able to predict the risk of injury, and we will be able to find innovations that will make our athletes safer while maintaining a high level of quality of play,” said Jeff Crandall, chairman of the NFL’s engineering committee. “
The NFL and Amazon have a lot of resources available. But it’s hard to predict injuries, especially in chaotic sports like rugby. “It’s the Holy Grail. Everyone wants to do it, and no one can do it. Zachary Binney, an epidemiologist and consultant who has worked with the NFL, said: “I was skeptical until I saw the results. “Predicting injuries is challenging because there are many factors that can cause damage. “It’s just a very difficult question,” Binney said. “
The Amazon Web Services Partnership will try to bridge the gap with league-level data from the NFL’s “next generation” statistics, which captures location data for each player for each player hundreds of times per minute through a chip on its pads. It also includes video footage of the game, information about the playing field and environmental factors, and anonymous player injury data, the NFL reported. Binney said it does not collect data on how body parts hit the ground or other participants, which is a limitation. However, it can be detailed to see how and at what speed the players play. The aim is to find out if any common elements in football are more likely than others to cause any harm.
“When the wide receiver moves quickly and turns, you may see what happens and maybe be able to pick something out of it,” Binney said. “League-level data includeonly only some measures of player activity. Each team has more detailed data on their athletes, often tracking all things such as heart rate, fatigue, and other metrics, all of which can lead to the risk of injury to a particular athlete. Other risk factors for injury in rugby include flexibility, injury history, strength and physical composition. However, much of the player-specific data stays at the team level to avoid providing potentially useful information about the performance of their players to opponents.
According to an email sent to The Verge by an NFL spokesman, player health data is not included in the injury prediction program. This may affect its predictive power. “It’s going to be really interesting. I don’t know what the impact might be, and I can’t imagine they have an impact. Binney said.
Even without more detailed information, the league has data from athletes from all 32 teams, providing them with more opportunities for collaboration. “You lost some details of the data, but you increased the sample size,” says Mr Binney. “
In the past, NFL employees manually checked hundreds of hours of game footage and helmet impacts to identify injuries and made changes ,such as updating startup rules, to prevent such situations. Binney speculates that the project may bring other changes, but any information they can collect may have other value.
If the NFL’s efforts in injury prediction and prevention prove effective, they can also provide a road map for other sports. Binney says it’s a positive step. “I’m happy to see it happen, although I’m wary of how much we might learn from it. “