“Citius, altius, fortius”, or “faster, higher, stronger” as the Olympic motto goes. As humans continue to push the limits of physical performance, how are emerging technologies playing a role ? Do AI and Big Data hold the key to continuing the trend ?
We’ve been playing sports for a long time and yet, somehow, the fastest swimmers continue to swim faster, strongest weightlifters stronger, and fastest runners still faster. Yet we humans haven’t appreciably evolved since the invention of sport, so how is it that these records continue to be broken? Naturally, there are several factors at play, ranging from improvements to equipment, breakthroughs in nutrition and exercise science, and sometimes even changes to the rules of a particular sport, but today we’re going to examine another factor at play: digital technology.
A good jumping-off point is David Epstein’s TED talk, and it’s a bit of a bombshell : athletes aren’t getting bigger, faster, stronger. At least, not really…
New artificial tracks, new shoes, new swimsuits… Our apparent ability to go push back the human body’s physical limits has more to do with progress in materials science than with biomechanics per se. Could the same be true with computers, Big Data and AI ?
In the world of professional sports, with the increasing crackdown against pharmacological doping and the proliferation of high-tech materials, the old methods of pulverizing world records are yielding diminishing returns.
Unless you’ve been living under a rock for the past couple of years, you’ve no doubt heard about the AI frenzy that is sweeping the globe. Athletes don’t live under rocks, and so naturally have turned towards this new and promising Eldorado for that additional, record-breaking edge.
Two such applications of AI to professional sports have caught our attention.
Customizing the workout
Companies like WHOOP are outfitting professional athletes with high-end wearable devices that sample biomechanical data at high frequency in order to improve recovery. The general idea is that certain patterns of movement correspond to certain types of physical activity (e.g. running, weight-lifting, or swimming), which in turn require different recovery regimens. Its watches are already being used by MLB players, which authorized the use of the device during games, and the NFS, with whom WHOOP has just finalized a deal.
WHOOP’s hardware can be distinguished from consumer-grade wearables by its sampling frequency : 100 Hz, which works out to about 100 megabytes per day. Compare this with the Apple watch, which clocks in at .01 Hz (i.e.: one sample every minute). It’s plain to see why Kevin Durant, the NFLPA and Russell Okung all joined the company’s investment club. WHOOP has already raised $25M last March through this and other VC firms.
Other companies such as the Peak Performance Project (P3), a training center and research lab in California, are pairing cutting-edge motion-capture technology with AI to create custom training regimens, tailored to individual athletes.
The hardware is impressive : ten cameras provide 3D motion analysis system and feed thousands of data-points per second into its server farm for online analysis.
As P3 states on its website, “P3 operates with the knowledge that everybody is constructed uniquely. No two athletes have the same biomechanical make-up, and as a result, no two athletes have the same physical strengths and weaknesses. The P3 sports science assessment is designed tease out these unique physical intricacies with the goal of developing custom exercise programs tailored to your individual needs.”
Safer sports thanks to AI
Fundamentally, there are two way to increase an athlete’s performance. The first and most straightforward is to increase the physical capabilities of the human body, but strange as it may sound, human capacity isn’t always the limiting factor in a sport. This is particularly evident in motor sports, where the human driver could easily push further, if only it weren’t prohibitively dangerous to do so.
Enter Argo AI, which can detect car malfunctions in real-time, and give their pilots a chance to respond.
The startup has been working with the Ford Motor Company for the past 5 years, burning through an impressive $1 billion runway (that’s billion with a B). Initially, the company was developing self-driving cars, but it’s now also expanding its operations into a sector called predictive maintenance. It’s arguably the first to expand this sector into the world of auto racing.
Of course, injuries and accidents are rampant in many other sports too. With the TBI epidemic in American football to the heaps of sprains, torn ligaments and stress-fractures in football and basketball, the millions paid each year to professional athletes are beginning to look like hazard pay.
Fortunately, predictive maintenance applies to people as well. In baseball for instance, shoulder and elbow injuries are represent the overwhelming majority of cases handled by sports doctors. This is especially the case for pitchers, who are prone to overusing their strong arm. To nip this in the bud, many teams are opting to use digital modeling to determine which kinds of plays most contribute to elbow and arm injuries, TechPulp has learned, though details are sparse. In football, a similar methodology is being used for early detection of concussions in players. It’s likely this advance will continue to align with the NFL’s recently established concussion protocols, and that details will progressively become public.