It’s not easy to be healthy. And it’s even harder to be healthy at work, where chances are–despite the vogue for standing desks and the like–you’re parked in a chair for most of the day, focused on a screen. The average workweek, by one recent measure, is now 47 hours and counting. By and large, more time spent at the office means more time hunched over a computer, probably eating lunch at a desk. Stress–which has been linked to increased risk of stroke, heart disease, and other health issues–is just about inevitable as a result.
Knowing how to be healthy, though, isn’t the biggest problem–many of us are better informed than ever about how to make healthy choices. And while federal guidelines may shift a bit over time, the basics–like eating more whole grains and vegetables and fewer processed foods, or moving more and sitting less–remain basically the same. Putting that knowledge into practice is the real challenge.
This isn’t lost on employers, many of which have already come to terms with the research pointing to correlations between productivity and healthy choices–from physical activity to good eating habits. More and more, you don’t have to work at Google (as I do) to gain access to an employer-provided gym or fitness stipends or even free healthy meals at work. But there are already signs that businesses are looking past those low-tech options to explore another way to boost their employees’ health and wellness: machine learning.
Broadly speaking, the biggest potential advantage to bringing in machine learning for on-the-job health is that it’s all about putting data we already have into action in real time. Fitness trackers do that to a certain extent already, and many of them are designed to use that data to help modify users’ behavior (though some are skeptical as to how successfully).
But not all tools based on machine learning are wearables, and many can still give you in-the-moment nudges in subtle, helpful ways (that don’t make you feel bad about yourself). A recent one that we’ve built here at Google is the Goals feature of Google Calendar, released last month and meant to help people find time for their goals–whether that’s going to the gym or wrapping up creative projects on schedule. After answering a few simple questions like “How often”? and “Best time?” Google Calendar finds the optimal window in your schedule and automatically reschedules if a conflict comes up.
Then there are innovations that are based on physical sensors but don’t come in the form of a watch and are purpose-built for the workplace. Smart furniture like the Stir Kinetic Desk, for example, calculates how much you sit and stand and based on goals you input and reminds you when it’s time to change your position. The Axia Smart Chair monitors your posture through sensors and adjusts accordingly. This Hyperchair adjusts to your body temperature to keep you warm or cool you down, and even has built-in Wi-Fi.
Of course, it’s easy to dismiss these as gadgets, add-ons, and widgets that merely tweak our working lives rather than revolutionize them. They might even make us act differently but not truly become healthier. And so far, anyway, that may be true. But psychologists (and most people, for that matter) have long known that shooting for dramatic behavioral changes can often lead to misses or just temporary gains.
Behavioral economics tells us that pre-committing to goals helps us make positive changes we can actually stick with. And machine-learning technology–which is evolving fast–is already pretty good at helping us pre-commit to manageable habit changes. (In fact, that’s arguably what most of it right now is best at.)
What if your software knew you get hungry around noon each day, so at 11:30 a.m., it proactively suggested a few healthy options from Sprig, Munchery, or Maple? If the game-time decision component was removed from choosing your lunch, machine-learning– equipped software could help you make choices before you’re starving and most likely to demolish a bag of chips rather than find your way to the nearest salad.
You might not even realize that machine learning is at your workplace already. Some offices set room temperatures to adjust automatically, knowing how that can impact productivity. The strength of light in each room may be programmed to change throughout the day.
Again, these may seem like small adjustments that are ultimately tangential to the sorts of behaviors we’re most likely to associate with better health: Either you do 20 minutes of cardio four days a week or you don’t. But focusing strictly on those habits–often the ones we try (and too often fail) to squeeze in outside of work–may risk missing other habit changes we can make during the workday, and the ways technology can help us make them.
More and more, machine learning won’t just help create a healthier workplace by fiddling around the margins. It will be at the core of a healthy, energized, and productive workforce. By learning and responding to our habits and desires, machine-learning technology has the potential to fundamentally change the way we work.
But there’s no guarantee that it actually will. In order for these changes to come together, the fragmented Internet of Things ecosystem will need to cohere. Today, data exists in silos, which limits its usefulness and requires individual users to extract meaning from ever greater heaps of data. If we can consolidate these various data streams in an easy-to-use software layer, all workplace devices–desks, chairs, wearables, software, not to mention the innovations we haven’t even thought to invent yet–will start working together to make healthier and more productive, even while we’re at the office and have our minds on other things.
Robin Mestre is principal strategic advisor at Google for Work.