Most of us fitness enthusiasts own a fitness activity tracker by now. But if somehow you haven’t fallen into the craze yet but you are tempted in buying one, how do you choose which one is right for you? You could choose by design, which one you think looks better on you, you do have to wear this all day after all. Alternatively, you could choose by brand, we all have our favourites, which could also mean compatibility to your phone or computer. Otherwise, if you are very serious about tuning your fitness and very interested in the data you collect, you could choose by function and accuracy. Or could you? There are no complete studies comparing all of fitness trackers out there and the ones that exist use different methods of measurement and information analysis meaning they cannot be, strictly speaking, truly compared.
So instead the real question might be, collectively are they worth their money? How much of their data is true? Can we trust technology wearables to make conclusions on our fitness and health?
There are many articles on pros and cons of each activity tracker, feature and price comparisons and even individual product reviews. As a result, this article is concerned with general research data published by scientists in peer-reviewed publications on function and accuracy. Actual scientific information and testing might shed some light into how you should use these trackers but more interestingly give you an understanding on where such devices work and fail. How much of their data is true?
Good news! Some research studies concluded that in general most activity trackers tend to, most of the time, under-estimate the energy expenditure. And there even might be an explanation of why you still feel tired in the morning as the total sleep time and efficiency was found to be over-estimated while the wake up time was under-estimated.
When it comes to step counting, hip worn trackers generally performed better than the ones worn at the wrist. Step counting appeared to be accurate with the higher error, most of the time, being an under-estimation of steps taken. However, the distance covered was over-estimated at slower speeds and under-estimated at faster speeds.
Fitness trackers generally over-estimated moderate to vigorous activity times. Thankfully though, most fitness trackers measured heart rate fairly accurately. Hopefully, future smart trackers will learn from the wearer’s behaviour and identify more correctly when and what the person is doing especially for things that are difficult to distinguish at the moment such as if the person is asleep, napping, or resting.
Depending on the brand, model and what type of user you are your fitness activity tracker could be both over and under-estimating your efforts. The main take away from all research studies on the matter is: use your fitness tracker as an estimation. A general indication of how you did that day compared to the previous one and not as actual data. While this might come as no surprise to some, there are others that reportedly eat, for example, as many kcals as their fitness tracker told them they burned. Take your fitness tracker data into consideration in a comparative way, did you do better today than yesterday, and not as an actual statement on your activities.
Would I recommend a fitness tracker? I personally owned a few different brands and found the Fitbit ones, especially the Fitbit Surge, to be the best in all categories function, design, wear ability etc. I even forgive its massive size on my tiny wrist. However, if you are truly serious about getting and staying fit and eating healthy you don’t really need one to achieve your efforts. If you think it will motivate you or help you, especially at the begging, then go ahead and buy the one that suits you but they are not a must in order to be and get fit. Especially since their data is only suggestive.
How to improve your fitness tracker’s functionality and accuracy
Researchers K. R. Evenson, M. M. Goto and R. D. Furberg published a list of strategies in order to improve the functionality and accuracy of your tracker in terms of steps, distance, physical activity, energy expenditure and sleep. Admittedly, most of these are fairly straightforward and expected but nevertheless worth mentioning as advice from the experts. These researchers spent hours reading and reviewing data of their own and of other researchers’ work on fitness activity trackers. I have copied their instructions below for you to read under the terms of the Creative Commons Attribution 4.0 International Licence (http://creativecommons.org/licenses/by/4.0/). Some footnotes, web links and extra examples were deleted to keep this list short and to the point. Bear in mind that these options might not be available on all fitness trackers. Fitbit and Jawbone are used as examples here as the authors identified them to be the major market players in April 2013-March 2014. While the market share might not be the same anymore, these examples still hold.
1. “Wear the tracker in the same position every day – While wearing the activity tracker in the same position daily may be obvious for the wrist-based trackers, those worn on a pocket, bra, or hip could vary in accuracy depending on location. Trackers are more accurate when worn close to the body.
2. Enter your details and sync – At initial set-up, users should accurately enter height, weight, gender, and age into the application and sync it to the tracker. For example, these characteristics, as well as heart rate if available, are used by the Fitbit to calculate energy expenditure. Related to this, if body weight meaningfully changes, then updating the tracker with the new weight would be important.
3. For wrist-worn trackers, indicate if wearing it on the dominant or non-dominant side – In the software set-up, indicate if possible whether the wrist-worn tracker is being worn on the dominant or non-dominant hand. For Jawbone, trackers worn on the non-dominant wrist may be more accurate, probably because the non-dominant hand is less active than the dominant hand, so it provides a better representation of overall body movement. Fitbit indicates that using the non-dominant hand setting increases sensitivity of step counting and can be used if the tracker is under counting steps.
4. Calibrate stride length – Calibrating stride length may improve distance measures. In our review, only one study indicated that this was performed. Fitbit indicates a default stride length is used otherwise, based on height and gender. Jawbone also provides information for calibration.
5. Use add-on features and obtain updates – Using add-on features and obtaining updates might become more important since future iterations of algorithms to calculate physical activity or energy expenditure may use new features, such as heart rate and respiration. For example, Fitbit indicates that trackers with heart rate better recognise “active minutes” for physical activities that do not incorporate stepping, such as weight lifting or rowing.
6. Add more information via the diary or journal function – Providing information to the tracker on the specific physical activity being performed can help the tracker learn what activities look like for the individual. This is particularly important if the algorithms used by the activity tracker rely on machine learning techniques.
7. Interact with the sleep mode settings – Interacting with the sleep mode settings may help the tracker learn if the user is sleeping, napping, or awake. Fitbit indicates that the normal mode counts significant movements as being awake and is appropriate for most users, while the sensitive setting will record nearly all movements as time awake.” 
References and interesting reads:
1. “Systematic review of the validity and reliability of consumer-wearable activity trackers” https://doi.org/10.1186/s12966-015-0314-1
3. “Are Currently Available Wearable Devices for Activity Tracking and Heart Rate Monitoring Accurate, Precise, and Medically beneficial?” https://doi.org/10.4258/hir.2015.21.4.315
4. “A comparison of wearable fitness devices” https://doi.org/10.1186/s12889-016-3059-0
5. “In Bed with Technology: Challenges and Opportunities for Sleep Tracking” https://doi.org/10.1145/2838739.2838742