A step is a step right? One foot down is one step done. What could be simpler?
But when it comes to fitness trackers, it is actually a little more complex. Understanding why step count from your tracker does not always equal the actual number of steps you took today is critical in understanding the medical potential of these devices.
How does a fitness tracker work? Most of them rely on a simple sensor called an accelerometer or pedometer. When you move the sensor will record the extent of that movement. It often captures the direction and magnitude of the movement. The result of walking with a fitness tracker on your wrist is that accelerometer or pedometer records a lot of movement in the form of magnitude and directions. An electrical signal that looks like a wave is likely generated. The peaks likely correspond to when you took a step and swung your arms – recording a step. But what if you took a smaller step. What if you did not move your arms? When does the electrical signal recorded actually register as a step and when does it not?
But when it comes to fitness trackers, it is actually a little more complex. Understanding why step count from your tracker does not always equal the actual number of steps you took today is critical in understanding the medical potential of these devices.
How does a fitness tracker work? Most of them rely on a simple sensor called an accelerometer or pedometer. When you move the sensor will record the extent of that movement. It often captures the direction and magnitude of the movement. The result of walking with a fitness tracker on your wrist is that accelerometer or pedometer records a lot of movement in the form of magnitude and directions. An electrical signal that looks like a wave is likely generated. The peaks likely correspond to when you took a step and swung your arms – recording a step. But what if you took a smaller step. What if you did not move your arms? When does the electrical signal recorded actually register as a step and when does it not?
Image from: http://web.mit.edu/bhadley/Public/GlassAtBoozAllen/sensors.html
Its all up to mathematics and what cut off point or step threshold is set by the fitness tracker. The fitness tracker may try to ignore signal it considers noise, data that does not make sense or fit usual patterns. This signal processing is often unique to each tracker. There may be preset rules say that any signal within 10% of the average peak after processing is a step. But why not 5%. Why not 20%. Each company of fitness tracker has their own formula and cut off for what counts as a step – and often those formulas are secret.
So why does this matter? Now imagine you have two people each using different fitness trackers. You want to know who has been more active. The first person says she has been more active because her step count is 5% higher than the second person. But you think to yourself….is that 5% due to actual steps or because of a difference between how the two different fitness trackers calculate steps. The problem quickly becomes clear.
Without access to raw data, it is hard to know what step count really means. Even within the same line of device a company puts out – did the formula for step count change in the last year? It is nearly impossible to reproduce studies without the raw data or knowing how step count is formulated. It is analogous to trying to measure the temperature of a population where no one will agree to calibrate their thermometers or share what they are recording. Step counters clearly work well enough for everyday use and the measurements they offer often seem right. But to move from the wellness and into the medical realm, we need to move from ‘seeming right’ to ‘being right.’ We need access to the raw data and not constructed units like step counts. It’s a subtle difference but a critical step in moving these devices in the right direction.
(note: some concepts are more complex than explained here, but hopefully the broad theme and challenge are clear)
So why does this matter? Now imagine you have two people each using different fitness trackers. You want to know who has been more active. The first person says she has been more active because her step count is 5% higher than the second person. But you think to yourself….is that 5% due to actual steps or because of a difference between how the two different fitness trackers calculate steps. The problem quickly becomes clear.
Without access to raw data, it is hard to know what step count really means. Even within the same line of device a company puts out – did the formula for step count change in the last year? It is nearly impossible to reproduce studies without the raw data or knowing how step count is formulated. It is analogous to trying to measure the temperature of a population where no one will agree to calibrate their thermometers or share what they are recording. Step counters clearly work well enough for everyday use and the measurements they offer often seem right. But to move from the wellness and into the medical realm, we need to move from ‘seeming right’ to ‘being right.’ We need access to the raw data and not constructed units like step counts. It’s a subtle difference but a critical step in moving these devices in the right direction.
(note: some concepts are more complex than explained here, but hopefully the broad theme and challenge are clear)
RSS Feed