The step counter calculates steps from the x-axis, the y-axis, or the z-axis, depending on the change in acceleration of the axis whichever is the largest. If the changes in acceleration are too small, the step counter will discard them. Trackers use algorithms to analyze raw accelerometer data to determine what is counted as a step. Many characteristics of the data can be considered, including the amplitude and frequency of 3D accelerations and the patterns of episodes and pauses of steps.
Each company has different algorithms (usually proprietary) with different thresholds for what are ultimately counted as steps. Therefore, different trackers can give rank numbers for the same activity. Many research studies have analyzed the accuracy of the tracker. Studies often compare a variety of consumer trackers to some kind of gold standard.
Most people wear their fitness trackers tied to their wrists. Wrist-based devices need to make a lot of estimates to be able to count steps. Most fitness trackers do this using something called a three-axis accelerometer, a small module that detects (you guessed it) acceleration. Gyroscopes also help fitness trackers track orientation, and altimeters help your portable device monitor altitude.
Most trackers use three-axis accelerometers to monitor arm movement in all directions, which can be freely translated into steps. Most trackers will extract roughly the same data here, but where things start to differ is how this dataset is interpreted. That's where algorithms come into play. Different fitness companies get heart rate sensors from different companies, and some of those sensors aren't all at the same level.
In the future, you may be able to adjust your own tracker to customize the analysis of your movement patterns with the push of a button. All of this information is collected and analyzed to create an overall reading, and the more sensors your tracker has, the more accurate the data will be. Wearable has a great explanation about how fitness trackers calculate steps if you want to go deeper, but that's the general gist. Some people don't like a snug fit for their watch or bracelet, but if your fitness band falls off your wrist, you're likely to get false step counts.
Many trackers ask you to specify whether you are wearing the device in your dominant or non-dominant hand. Companies like Fitbit and Google are focusing their activity tracking around these metrics within the Fitbit app and Google Fit. Don't be scared because suddenly you're 5 km above or below your previous average, instead get used to the new tracker and adjust to the new baseline. Therefore, a wrist tracker may not measure the activity you do while your arms are still (riding a bicycle, carrying food, or pushing a stroller or lawn mower).
Ultimately, fitness devices are highly personal, functional, and often affordable pieces of technology that can actually track an incredible amount of data. Philippe Kahn is co-founder and CEO of FullPower, developer of MotionX, and explained to us how the “signal processing” procedure within a fitness tracker cleans the raw data that is collected. Bioimpedance sensors check the skin's resistance to a small electrical current and the four electrodes inside the UP3 activity tracker are clearly visible. To accumulate false steps, you can sit on a sofa and shake your wrist; stick your fitness tracker in the dryer; or connect your tracker to an electric saw blade, power drill, paint can shaker, ceiling fan, metronome, dog, bicycle wheel or hamster wheel.
As you may already know, it's hard to get two fitness trackers to agree on how much activity you've been doing in a day or what your heart rate really is. In a smart study, researchers provided participants with smartphone-based activity trackers that used an accelerometer and a gyroscope (for the speed of rotation around each axis). .
