4 Ways to Use the Training Data from Wearable Tech

“],”renderIntial”:true,”wordCount”:350}”>

The central question that sports researchers are grappling with these times is this: What the heck are we going to do with all this data? In endurance sports activities, we have progressed from heart level monitors and GPS watches to innovative biomechanical analysis, interior oxygen degrees, and steady glucose measurements, all exhibited on your wrist then immediately downloaded to your laptop. Crew sporting activities have gone through a related tech revolution. The resulting info is intriguing and ample, but is it basically handy?

A new paper in the Intercontinental Journal of Sports Physiology and Performance tackles this concern and provides an appealing framework for thinking about it, derived from the enterprise analytics literature. The paper will come from Kobe Houtmeyers and Arne Jaspers of KU Leuven in Belgium, alongside with Pedro Figueiredo of the Portuguese Soccer Federation’s Portugal Football Faculty.

Here’s their four-stage framework for information analytics, presented in get of both equally increasing complexity and growing value to the athlete or coach:

  • Descriptive: What transpired?
  • Diagnostic: Why did it transpire?
  • Predictive: What will happen?
  • Prescriptive: How do we make it occur?

Each and every phase builds on the former one particular, which indicates that the descriptive layer is the foundation for everything else. Is the data excellent more than enough? I’m pretty assured that a modern-day GPS view can correctly explain how far and how quickly I’ve run in schooling, which enables me to shift to the up coming phase and check out to diagnose whether a good or lousy race resulted from schooling far too considerably, too minimal, much too really hard, as well easy, and so on. In distinction, the heart fee facts I get from wrist sensors on sports watches is utter garbage (as verified by evaluating it to knowledge from upper body straps). It took me a when to realize that, and any insights I drew from that flawed details would of course have been meaningless and probably harming to my coaching.

Producing predictions is tougher (especially, as the saying goes, about the potential). Researchers in a range of sporting activities have experimented with to use machine discovering to comb via large sets of education information to forecast who’s at significant hazard of having hurt. For case in point, a study published previously this year by scientists at the University of Groningen in the Netherlands plugged 7 yrs of schooling and personal injury information from 74 competitive runners into an algorithm that parsed danger centered on possibly the earlier 7 times of jogging (with ten parameters for just about every working day, like the complete distance in distinctive education zones, perceived exertion, and length of cross-coaching) or the past three months (with 22 parameters for every 7 days). The ensuing design, like related kinds in other athletics, was significantly superior than a coin toss at predicting accidents, but not nevertheless good sufficient to foundation teaching selections on.

Prescriptive analytics, the holy grail for athletics scientists, is even a lot more elusive. A straightforward case in point that doesn’t demand any major computation is heart-fee variability (HRV), a proxy measure of pressure and recovery position that (as I discussed in a 2018 report) has been proposed as a everyday tutorial for deciding irrespective of whether to coach tricky or effortless. Even however the physiology makes sense, I have been skeptical of delegating vital training choices to an algorithm. Which is a false choice, though, in accordance to Houtmeyers and his colleagues. Prescriptive analytics delivers “decision guidance systems”: the algorithm is not changing the coach, but is delivering him or her with a different point of view which is not weighed down by the unavoidable cognitive biases that afflict human final decision-making.

Curiously, Marco Altini, a person of the leaders in building methods to HRV-guided coaching, posted a Twitter thread a couple months ago in which he mirrored on what has changed in the subject because my 2018 posting. Among the insights: the measuring know-how has enhanced, as has understanding about how and when to use it to get the most trustworthy data. That is critical for descriptive use. But even good knowledge does not promise excellent prescriptive tips. According to Altini, reports of HRV-guided education (like this one particular) have moved away from tweaking work out plans based on the vagaries of that morning’s studying, relying alternatively on for a longer time-term developments like functioning 7-working day averages. Even with those people caveats, I’d continue to view HRV as a supply of selection guidance instead than as a choice-maker.

One of the good reasons Houtmeyers’s paper appealed to me is that I spent a bunch of time imagining about these concerns in the course of my modern experiment with continual glucose monitoring. The four-stage framework aids explain my contemplating. It’s apparent that CGMs present terrific descriptive info and with some work, I imagine you can also get some superior diagnostic insights. But the revenue pitch, as you’d count on, is explicitly targeted on predictive and prescriptive promises: guiding you on what and when to try to eat in order to optimize effectiveness and recovery. Perhaps which is attainable, but I’m not however certain.

In fact, if there is just one uncomplicated information I choose absent from this paper, it is that description and analysis are not the same matter as prediction and prescription. The latter does not follow quickly from the former. As the information sets hold obtaining greater and larger-high quality, it appears to be inescapable that we’ll sooner or later achieve the issue when machine-learning algorithms can select up patterns and interactions that even highly experienced coaches may well miss out on. But that is a major leap, and information on its own—even “big” data—won’t get us there.


For more Sweat Science, sign up for me on Twitter and Fb, signal up for the e mail e-newsletter, and look at out my book Endure: Head, System, and the Curiously Elastic Restrictions of Human General performance.