Super Bowl AI: How NFL’s ‘Digital Athlete’ helps reduce concussions

The National Football League (NFL) has developed an artificial intelligence tool that uses TV imagery and sensors embedded in helmets, mouth guards and shoulder pads to try and reduce injuries in American football games.

The NFL “Digital Athlete”, developed with the help of Amazon Web Services, creates a digital replica of an NFL athlete in a virtual environment. Using machine learning and computer vision technology, it pinpoints impacts and injuries and helps find ways to make players safer.

“Having the computers understand how many times a player hits his helmet during the course of a game”, says Jeff Miller, NFL Executive Vice President, “[helps] find ways to reduce the amount of helmet contact”.

In this virtual environment, an infinite number of game scenarios and environmental situations can be run, “giving the ability to test out new safety equipment, test out rule changes and predict player injury events and recovery trajectories eventually”, says Dr Priya Ponnapalli, principal scientist at Amazon Machine Learning Solutions Lab.

Head trauma in NFL games has previously been linked with brain damage in ex-footballers. As a result, the risks of injuries to athletes participating in collision sports has been more prominently reported in recent years. 

A 2017 study of 111 brains obtained from deceased former NFL players found that 99 per cent had chronic traumatic encephalopathy, a neurodegenerative disease. “What we’ve shown, I would say pretty definitively, is the relationship between years of play and risk of the disease,” said Jesse Mez, Associate Professor of Neurology at Boston University.  Mez hopes that the NFL will make more data available for study, “right now no helmet sensor data [is] made available to any investigators at universities. I think that would be very valuable data for us to be able to implement”, he says. 

Ultimately, the more steps taken to reduce injuries can only be a good thing, “more players who are healthier, longer is good for everybody and that’s what our goal is”, says Miller.