AI and a Brave New Running World
AI is the inevitable future of athletics posed to revolutionize the endurance sporting industry. Here’s where we are, and what we can expect to see in the coming years.
Heading out the door? Read this article on the new Outside+ app available now on iOS devices for members! Download the app.
Once a SciFi plot, there is almost no sphere of our lives that Artificial Intelligence and machine learning has left untouched. From security intel to education to entertainment, AI, machine learning, and big data are posed to radically transform the economy, politics, the military, and our personal lives.
And so, you can be certain that the future of endurance sports is going to be revolutionized by AI in the coming years. Indeed it’s already rapidly changing a variety of sports as we’ve known them through algorithm-generated personalized training programs, smart clothing, multi-sensor wearable technology, and advanced performance analytics.
What is AI?
While tech companies and political leaders have entered into a battle to rule the big data sphere, most of us still remain in the dark about what AI even is. It’s a topic shrouded in complex jargon, muddied by fictional futuristic plot lines. But AI is a real thing, it’s here to stay, and it’s already having powerful implications in the realm of endurance sports.
Put (very) simply, AI refers to computerized machines that can learn, reason, and act independently, for themselves as if human. They can use information to make their own decisions when faced with new situations. Most advancements in the realm of AI that you hear about refer to “machine learning,” a category of algorithms that detect patterns in massive amounts of data (“big data”). The computer can then use that pattern to make predictions, for example what music you might like or, in the case of wearable AI like the multi-senor smart clothing, what injuries you could be prone to based on your gait analysis.
Where We’re At
Currently, AI and Big Data are set to dominate the booming sports industry. Globally, the industry generated 90 billion dollars in 2017. The North American sporting industry alone produced a revenue of 73.46 billion dollars in 2019 — and these numbers are predicted to keep growing. This is an industry only becoming more competitive by the year, and it’s set to become more so with the amalgamation of a variety of AI systems based on biomechanics and multi-sensor technology.
All sports are a game of numbers: Victory, defeats, and record-breaking performances, pitching velocity, free-throw percentages, vertical leap and myriad more metrics are measure and then translated into digestible and engaging information that allows experts to tell stories and make predictions, coaches to analyze performances, and managers to calculate budgets. Endurance athletes and coaches typically delve deeper into performance metrics like workout interval times, race splits, heart rate, VO2 max, energy output levels, symmetry, velocity, acceleration, cadence. Because of the use of AI systems, we know that competition outcomes and training success are the result of patterns in the numbers rather than chance. And so, AI technology is increasingly invaluable to endurance athletes, coaches, and the sporting industry at large.
AI-Generated Training Programs and Performance Predictions
Because AI systems are able to look at and reveal patterns in a variety of data on biomechanical and physiological factors that affect athletes’ performances — volume, speed, VO2 max, heart rate, and even sleeping patterns to name a few — they can help an individual create an optimal training program tailored to his or her specific physiology, strengths and weaknesses. This is often done by collecting data from single-sensor wearables, with multiple built-in technologies that can measure and analyze internal and external variables that affect an athlete’s performance and fitness progression.
Last October, PodiumRunner published two articles looking at how big data was revealing optimal training patterns and performance predictions for the marathon or other long distance races. One examined how big data is revealing what training patterns are most effective for optimal performance in longer races like the marathon. The other reported on a team of scientists from France and Finland who developed a way to predict an individual’s marathon time based on smart-watch data logged during six months of training. Incredibly, they were able to estimate thousands of runners’ race times within, on average, a 2% margin of the time they actually ran.
Though this isn’t currently an app, it may soon be. The study’s main author, Thorsten Emig, told journalist Richard Lovett that their team hoped to make it available by “the end of the year.” You can imagine that once these predictors become available to the masses, the rise an influx of AI-driven training programs available for runners to use as a sort of “iCoach.” While it’s unlikely that coaches and personal trainers will be completely replaced by AI technology, especially at more competitive levels in the sport, we can expect to see AI “coaching” become more integrated into endurance sports.
In fact, it’s already happening. Take, for example, the rise of AI-generated personalized training programs and coaching like AI Endurance, founded by Markus Rummel in 2020. It’s an app that uses AI-driven algorithms to put together individualized training plans for endurance athletes by using that specific athlete’s biodata to help them optimize workouts and race performances. The app works by connecting your profile with Strava or Garmin and then analyzing your athletic history and health data to calculate what will work best for you as an athlete and with your schedule.
Injury Prevention and Health Monitoring
Because AI tech is able to find patterns in various health aspects and gauge athlete movement, the metrics can help detect early warning signs of injury or other physical misalignments — or low energy levels — thus rescuing athletes from serious injuries, like stress fractures, muscular tears, or burn out.
For example, tech companies like Sensoria and Torq Labs are beginning to develop “smart clothing” that use multi-sensor wearable technology to track and analyze biomechanical motion, gait, and alignment to detect signs of physical weakness, misalignments, or poor running form. Sensoria’s smart socks are able to use multi sensors to detect cadence, footlanding, and impact forces, while Torq’s clothing and corresponding app work as a sort of “prehab” by showing athletes potential injuries they may be at risk for before they happen.
“Endurance athletes could particularly benefit from Torq Smart Clothing by being able to understand the impact their training environment actually has on their body and gait efficiency,” says Torq Labs CEO Julian Holtzman. “The only way to understand that is to adapt their training schedules based on their-own optimal performance. More explicitly, an endurance athlete will be able to understand the total load impact on their body from the entire week of training down to the granularity of the segments of cardio where we can see their form breaking down. The pure awareness that something is not as it should be, or could be improved upon, is the first step to making that a reality.”
On the cardiac health front specifically, Ambiotex’s smart shirt uses ECG sensor technology to measure stress, heart rate variability (HRV), and anaerobic threshold via an integrated chest strap sensor. The data can then be analyzed and layed out visually through the company’s app.
Some of these technologies still face the challenge of providing adequate accuracy balanced with convenience of battery life, size, weight and cost. And while scientists debate the ability to generalize biomechanical metrics to predict “optimal” or “pathological” parameters for individual bodies, the technologies and knowledge are rapidly evolving and advancing.
This integration of AI in sport tech isn’t going anywhere. As the tech improves with the development of more accurate and user-friendly sensors, larger sets of data, and more precise algorithms, this will likely become even more pervasive and necessary for competing at high levels.
Here’s how we can expect to see AI, machine learning, big data, and multi-sensor technology revolutionize the sport tech industry in the next few years.
Smarter, More Personal Fitness Trackers and Sensors
It’s probable that in the near future wearable technology will radically enhance personalized training programs by taking harder-to-measure biological data from individuals into account. Currently, smart fitness trackers like Apple Watches, FitBits, and Garmin devices are ubiquitous in running communities for their ability to measure heart rate and track things like running pace and distance, and even levels of stress.
Next we can expect upgraded versions of these fitness trackers able to A) gauge more accurate and relevant data and B) use that data to prescribe better training models based on that data. Already, companies like WHOOP, a Boston-based startup, are raising the stakes. WHOOP monitors strain, recovery, and sleep cycles using algorithms based on HRV and four other variables tracked 100 times per second. Commercial AI wearables may soon be able to collect increasing amounts of biological information such as blood-glucose levels, sleeping patterns, heart rate cycles, hormonal fluctuations and more to create precise and flexible training programs able to recalibrate based on biological feedback.
An example is this algorithm-driven wearable skin patch for runners — currently just a prototype — that can measure rate of fluid and electrolyte loss in real time during a race and could eventually allow runners to fine-tune their hydration and nutritional plan during a race. The idea is to use the algorithms to get critical information to athletes in easy-to-use form via a mobile app that can interpret the results. This is likely the beginning of more low-cost wearable sensing devices that can improve the accessibility of physiological insights available to athletes, coaches, and trainers to inform and optimize performance strategies or detect injuries well before they can occur.
This includes technology applied to the body externally ( Apple Watch, Torq Smart Clothing) and technology applied to the body internally (Neuralink, pacemaker etc.). Broadly speaking, the many tools that fall under the AI umbrella, particularly Machine Learning, can pave the way for endurance athletes to start asking better questions about their performance and provide them with the tools to find the answers.
Enhanced Spectator Experience
Expect to witness a large-scale use of AI at major athletic events as soon as this summer. It’s expected to play a large role in the 2020 Tokyo Olympics. The tech giant Intel has partnered with the International Olympic Committee (IOC) and the Tokyo Organizing Committee of the Olympic Games to provide tools such as “3D Athlete Tracking” (3DAT) for spectators in Games venues. This will essentially use cameras with AI and computer vision to show fans almost real-time data during events, for example track races. This will provide a biomechanical on-screen analysis of factors like athlete speed, who is leading a race, and distance traveled / remaining in a race. Additionally, Virtual Reality, Augmented Reality, and Mixed Reality are also going to be used to enhance the experience for spectators in certain athletic events. Fans of the Olympic Games will be able to get VR headsets that are being promoted as a more immersive spectating experience than simply watching on TV.
As the first direct area of consumer use of this technology, the Tokyo Games is set to be a technological spectacle heralding a brave new world of athletics born from the marriage of artificial intelligence and sports tech. (That is, of course, if the Games happen at all.)
Ethical Questions and Concerns
As thrilling as the integration of AI tech into sports is, it comes with a host of philosophical, privacy, legal, and ethical concerns. If we can enhance our biology with artificial intelligence by breaking bodies down into quantifiable, predictable parts and optimizing those parts for better athletic performance, will competition begin to seem superficial? Is it ethical for big data companies to be able to buy, analyze, and profit off of our personal biological data? Are we willing to give up our privacy for enhanced training and health benefits? And what about sportsmanship ethics: If certain athletes are given access to performance enhancing AI technology, does that give them an unfair advantage over their competition?
These are all questions that will be debated relentlessly in the next decade as AI amps up its domination of sports tech. But one thing we know, whether we like it or not, is that it’s the present and future reality.