Putting Your Fitness Tech Data to Work
The avalanche of data generated by fitness tech has science zeroing in on some surprising performance recommendations.
Every day, as hundreds of thousands of athletes around the world fire up their Strava apps, Nike+ FuelBands, Fitbit Flexes, and other wearable-tech devices, they produce a mind-boggling amount of data.
In 2013, Strava users recorded 53.3 million runs and rides totaling 905,408,836 miles. In the Fuel-Band’s first year on the market, Nike claims that users generated enough kinetic energy to light up more than 6,700 homes. Even bike-sharing services are amassing data. B-cycle, which runs programs in 31 cities, reports that, in 2013, its 3,813 bikes clocked 1,532,836 miles over 719,641 trips. And the International Mountain Bike Association’s (IMBA) crowdsourced trail-finder site, MTBproject.com, contains 21,328 miles of GPS-mapped trails, with hundreds of miles of new routes being added each month.
Now that vast amount of back-end data is being used to effect real-world change. It’s already driving policy innovation: Oregon’s department of transportation has purchased Strava usage stats to improve its cycling infrastructure, right down to considering how often street cleaners should sweep bike routes in cities like Corvallis. In Arizona, IMBA tapped trail-use data to work with the Forest Service to allow bikes on several formerly illegal but well-known singletrack routes around Sedona. And the Outdoor Alliance’s exhaustive visitation stats helped federal land managers expand the 2012 Colorado Roadless Ruling from an initial 500,000 acres to 1.2 million.
But perhaps the greatest impact is happening in the health and fitness world, as researchers leverage all those bits that chronicle our routes, distances, times, and heart rates to fine-tune formulas for peak performance. Jawbone, the maker of the Up activity tracker, has found that among its thousands of users worldwide, jet lag from a coast-to-coast trip usually upsets sleep patterns for at least five days. Basis, maker of a wristwatch-style fitness and sleep tracker, is working with the University of California at San Francisco and others on sleep studies, including one that mined user info to prove that one of the most effective predictors of quality sleep is a consistent bedtime.
A vast amount of back-end data is being used to effect real-world change. And it’s already driving policy innovation.
Companies are also using the data on daily habits to make concrete training prescriptions. Jawbone has found that Monday is the most popular day for workouts. (Perhaps unsurprisingly, Sunday is the least.) Strava users seem to go hardest and fastest on Wednesday, Friday, and Saturday. The takeaway? Don’t plan your high-intensity interval rides for a Thursday when, for whatever reason, the data tells us you won’t be as into it.
Colorado Springs–based Carmichael Training System regularly draws on data points culled from its work with thousands of cyclists, runners, and triathletes to guide its coaching strategies. Among the nuggets learned from years of GPS, heart-rate, and power-meter data files: Contrary to popular assumptions, mountain biking is as effective at building competition-level fitness as road riding. Those who follow its training programs closely experience fewer injuries than those who don’t. And athletes can put up maximum power numbers for as many as three consecutive days with no loss of output—despite their own perceptions that they’re losing strength.
Ten years ago, this type of data was the exclusive domain of elite athletes and a smattering of bioscience labs. “But no one looked at the data to learn from it,” says Gear Fisher, founder of TrainingPeaks, a Boulder, Colorado, online coaching platform. (TrainingPeaks’ integrative training plans are also published on Outside Online.) “They used the technology to chart real-time performance, and then they forgot about it.”
That’s why this summer, Fisher’s company is rolling out an update of its WKO+ software, which Fisher believes is one of the most accurate exercise-modeling programs ever. “We’ll be able to predict performance based on as little as one workout,” he says. The data comes from numbers collected through TrainingPeaks.com, which is used by thousands of coaches to manage tens of thousands of runners, triathletes, and cyclists.
Looking at all those past performances, the company will predict results for new customers. “You’ll be able to see what you’re capable of at your current level of fitness,” says Fisher, “and soon you’ll also see what you need to do to reach a specific goal, like a 13-hour finish at Ironman Florida.” That’s right—not just any Ironman, but that particular Ironman. “You may not want to do what’s required to get there,” Fisher concedes. “But we can tell you if you can.”
Looking ahead, Strava cofounder Michael Horvath sees a day when user data can help race directors design courses that challenge—but don’t destroy—participants. “We’d be able to tell how much climbing is too much from completion rates and where people quit a race,” says Horvath. He even sees it helping gear manufacturers. “Users can track the number of miles they’ve put on their running shoes before they swap in a new pair,” he says, “and from the aggregate data, we’d know how many miles runners can get from that specific model.”
The rub, of course, is that people have to actually wear the devices and upload their results. In addition, the sample size, while enormous in scientific terms, is nonetheless self-selecting: active users of wearable tech. “The best you can say about the data is that it can be used to draw useful conclusions about the people who are using each app, like Strava,” says Yuri Feito, an assistant professor of exercise science at Kennesaw State University in Maryland. Still, says Feito, “Statistically, the level of information involved with Strava dwarfs anything that a research lab could pull together on a survey of cyclists. That shouldn’t be ignored.”
Stats from the Data Revolution:
- Increased likelihood of achieving a fitness goal when logging training and following a plan: 100 percent. (TrainingPeaks)
- Fitness-program success rate among participants who shared their workouts via social media: 85 percent. (411Fit)
- Extra weight lost in a month when logging an additional three days of food-diary entries: a third of a pound. (411Fit)
- Most common cross-training exercise for runners: swimming. (Jawbone)
- Most popular activity among females in Los Angeles: hiking. (Jawbone)
- Improvement in performance when working out with a coach: 10 to 20 percent. (TrainingPeaks)
- Average length of bike rides in 2013: 20.5 miles. (Strava)
- Average length of runs in 2013: 4.7 miles. (Strava)
- Additional sleep per night enjoyed by climbers versus other Jawbone users: 8 minutes. (Jawbone)
- Most active week in 2013 for cycling and running: August 25 to 31. (Strava)