Study: Smartphones Just As Good As Fitness Trackers For Counting Steps

jfruh writes While dedicated fitness trackers that you wear around your wrist have any number of functions, many people are focused on a single metric: counting steps, which serves as a proxy for determining how active you are. But a recent study from the University of Pennsylvania showed that if that's mainly what you want out of a fitness tracker, then you almost certainly have a device in your pocket that can do the same thing as well if not better: your smartphone.

Comments

Bad Summary: Looking at the graphs in the study

[Chrisq]

Looking at the graphs in the study a more accurate report would be:

Of the dedicated devices the "Fitbit One" and the "Fitbit Zip" where super accurate, but the average for dedicated devices was brought down by the abysmal performance of the "Nike Fuelband". The various apps tested gave a reasonable performance, as did the other dedicated devices tested

Re:Bad Summary: Looking at the graphs in the study

[swillden]

"Fitbit only usable device, Nike Fuelband total shit".

I think they're all quite usable. In this space utility is not dependent on accuracy, for most people, because for most people it really doesn't matter if you get an extremely accurate step count. What matters is that you get a reasonable estimate of your activity level, as compared with you activity level measured on other days.

Fitbit tries to estimate calories burned from your walking, and if that were at all accurate, it could arguably be important to get an accurate step count. But it's not, because people are too different.

For most of a year I performed very detailed tracking of my caloric intake vs expenditure, including the use of a Fitbit step tracker and a cycling power meter, and plotted those against my weight gain/loss. What I found, once I had determined by base metabolic rate, was that estimating caloric expenditure with the power meter is highly accurate, but Fitbit's estimate of my caloric expenditure due to walking consistently overreported by about 30%.

That's not because Fitbit is doing anything wrong; their pedometers are quite accurate and I'm sure they're using reasonable models for estimating energy use based on the walker's weight and height. The fact that the overreporting rate was so consistent even though my body weight changed by almost 15% supports this. A friend who did the same experiment for a while found that for him the Fitbit-reported caloric expenditure only slightly overreported. He and I are different; perhaps his walking style is significantly less efficient than mine, or perhaps it's something in the composition of our diets (the human body is an imperfect heat engine; different sorts of foods are converted with varying efficiencies, and even the combinations matter).

Then I tried using my phone to measure my steps. It reported far fewer steps than my Fitbit did every day, which concerned me, so I did some tests. I walked some measured courses with both and counted my steps. The Fitbit was darned near perfect, being off by only a step or two in thousands -- and it's entirely possible that those errors were mine, not the device's, since I was counting mentally rather than using a counting device. The phone, was bad... but in a consistent way. In fact, I found that, for me, applying Fitbit's formula (using their web site) but my phone's step counts actually provided a much better estimate of caloric expenditure, when calibrated against weight change.

If you're actually trying to convert steps to calories, I think inter-person variability is a bigger factor than device accuracy, and if you're anal enough to actually try to calibrate the system, your calibration for the former will easily take care of the latter -- as long as you consistently use the same device.

However most people don't, and won't ever, do the controlled experimentation and analysis to calibrate themselves. So what most people really use a pedometer for is to (a) track relative physical activity level over time and (b) set and work towards goals of increasing activity. If what matters is relative activity, then an inaccurate but reasonably consistent device is just as good as a perfectly-accurate device.

And you probably already have a phone, which is a reasonably-consistent pedometer. If you already have a Nike Fuelband, that works, too.

(And now I'm waiting to see if my AC stalker shows up and crapfloods responses to this post.)

Re:But the price...

[gstoddart]

Obviously you've never seen the weather-proof armbands people use for jogging with their cell phone ... put in the headphones, put it in the sealed case, strap it to your arm and go do your thing.

This is literally a solved problem, and has been for a bunch of years.

The accessories makers have been all over covering the running crowd for a LONG time.