Calorie tracking - is the band really accurate at all?

[Dima Habenko]

So MS has been claiming in their promotional materials that they worked for 3 years on the band and were able to create more accurate and personal energy tracking based on various exercises where traditional step trackers like FitBit would fail.
I was originally very excited as my FitBit was very limiting to only exercises like walking or running and anything else I had to enter manually by selecting the activity and intensity where FitBit would then assign expended calories based on their generic calculation tables.

But when I started using MS Band, I found that the energy tracking was working very different whether I was in exercise mode or not. The very same routines were registering differently depending on if I remembered to start exercise mode on the band or not. Apparently, the band relies a lot on HRM tracking for exercises that do not involve steps or where step tracking becomes unreliable. And where HRM tracking fails, so does the band algorithms.

Remember that famous bug in FitBit where it was tracking your car ride as some ridiculously intensive run because it was too sensitive to car vibration and registered it as steps? Unfortunately, what I found, is that while band is pretty conservative at filtering out false steps (too good for it's own sake sometimes), it's reliance on HRM fails it in car rides just the same.

I decided to track how much energy I spent by sitting in 1 hour traffic on my morning commute. And since the most accurate way to do that, according to MS, is by entering into exercise mode, so I did. By the time I reached office I spent more then 1000 calories!!! And also, according to the band, my average heart rate was close to 180 the whole ride with some peaks over 200!!! That is, except for the times when my hand was off the steering wheel. Because then my heart rate was a more normal 75-80 bpm. The band HRM was too sensitive to the micro vibrations of the steering wheel on the road and even though all other sensors (accelerometers, gyrometers, etc.) should have been telling it that I don't move (relative to my car) and therefore, should not spend my energy, the HRM was showing a cardiac arrest levels and the happy band was counting calories like crazy.
So the question is, has the Band made any improvements in the calorie tracking accuracy, or did it simply exchanged one unreliable tracking mechanism with new one, as unreliable as the old?

 

[eyecrispy]

Lol wow. You must be a crazy driver! Sounds like they have much tweaking to do. Thankfully, the hardware and data is all there. They just need to change the way it calculates results for us. They need to turn that around quickly before someone else does it and they lose the momentum they have right now.

 

[gadgetrants]

Hmmm, I was going to suggest you really look into (a) other commute methods and (b) anger management, but it sounds like the "vibration" theory is pretty compelling! I guess the idea that you have road-rage issues doesn't account for the ridiculously high HR. LOL

-Matt

 

[smurfalarm]

You're expecting a level of accuracy that is infeasible for something you wear on your wrist for everyday activities. If it's not measuring your CO2 output, then it's not a calorimeter and it can't accurately track calories. The Band is technically constrained to making estimates based on the inputs you give it and other correlated measurements. You told it you were exercising, and it (incorrectly) measured you as having a really high heart rate without taking any steps... which is exactly identical to the measurements it would get if you were in the middle of an intense cycling workout. It was reasonable given those inaccurate inputs that you were burning a lot of calories. Garbage in, garbage out.

I would be annoyed at the bad heart rate measurement, but at the end of the day, the really inaccurate calorie count is an easily avoidable problem. If you want to improve calorie tracking, don't tell the Band you are exercising when you know you're not actually exercising. I suggest you delete that workout from your activity history. After that, it should be significantly more accurate. Again though, it's a heart rate monitor, not a calorimeter. Even if you accurately report your workouts and the heart rate monitoring is accurate, the calorie count will still be an estimate, not a measurement.

 

[DroidUser42]

While I don't know what the Band uses for calculating calories, I suspect it's mostly heart rate. Bad HR measurement will throw off everything. And as you can see from other threads, the HR thing seems to be hit or miss. It works fine for some people, not so great for others.

Now, something to keep in mind for HR - during run or exercise, the HR monitor runs full time. Under normal conditions, it only runs 1 minute and shuts off for 9. So it's not going to do a good job of tracking exercise or changes in HR.

But I've also seen rate changes from normal and run. I'll have to explore that to see if it's just slow in picking up the change from driving and walking or if it's something more.

 

[dainla]

​I tend not to lie to my band and then I don't get strange information.

 

[Blacklac]

I promise you didn't burn 1000 calories sitting in your car on a commute to work. Lol

 

[gadgetrants]

I promise you didn't burn 1000 calories sitting in your car on a commute to work. Lol

Case in point, though. I run a few times a week (45 minutes) and I also teach college students 4 times a week (1:15 per class). I'm still scratching my head that the Band tells me I burn about 675 calories doing either activity. I'd never thought that lecturing and jogging were equally strenuous! LOL (Insert here jokes and sarcastic comments about the stress of teaching today's college students, or how spazzy an instructor I must be.)

But more seriously, I should note that I'm gauging my calorie burn while teaching using the "Exercise" (dumbbell icon) mode. I should maybe try to gauge it in Run mode, and possibly also just leave the Band in its default mode, to see how the estimate varies. I get the feeling that doing it in Exercise mode is overestimating...a bit.

-Matt

 

[lostlogik]

But more seriously, I should note that I'm gauging my calorie burn while teaching using the "Exercise" (dumbbell icon) mode. I should maybe try to gauge it in Run mode, and possibly also just leave the Band in its default mode, to see how the estimate varies. I get the feeling that doing it in Exercise mode is overestimating...a bit.

-Matt

I think by putting it into one of the modes it is expecting you to be doing something strenuous and therefore using different algorithm for calculating calories. I've left mine in default and then when checking though the "home" screen see that today I've burnt 1964 calories, which seems about right for the (non-exercising) day I've had.

 

[RiftPoint]

I think by putting it into one of the modes it is expecting you to be doing something strenuous and therefore using different algorithm for calculating calories. I've left mine in default and then when checking though the "home" screen see that today I've burnt 1964 calories, which seems about right for the (non-exercising) day I've had.

From what I've been reading recently, it seems like this might be the case. Apparently calorie burn can be essentially be estimated fairly accurately with a cardio activity because increased heart rate is based on system increase in oxygen need, but with something like lifting weights or other strength training exercises you can't use the same assumptions because you're involving your body is a more isolated way.

This correlates to why my 50min workout with weights only had me burning ~256 calories, whereas an hour long run had me burning ~752 calories. It makes me wonder if the band is programmed in such a way that if you enable something like running or exercising modes that its going to assume that you're doing what you told it you were and adjust it's sensitivity accordingly. For example, if you turn on exercise mode and then just walk around does the band assume that it's inputs are wrong because it's in that mode? If it did, then it would keep altering it's input thresholds until they yielded something that made more sense it's programming assumptions...

 

[mtalinm]

it just told me I burned 500+ calories during a half hour on the treadmill. now I'm going to do what I do with other fitness devices and tell it that I weigh half as much as I really do. that usually fixes it.

not like MS Health does weight tracking like MFP anyway, so no gr8 loss

 

[DroidUser42]

For example, if you turn on exercise mode and then just walk around does the band assume that it's inputs are wrong because it's in that mode? If it did, then it would keep altering it's input thresholds until they yielded something that made more sense it's programming assumptions...

Interesting idea. I kinda wonder if my test of putting it into "run" mode and then walking around is confusing it. When I walk the arms hang down. Jogging/running, I'd think they'd be in a "L" shape. And I have found that it's the movement that seems to mess with it.

 

[happydude602]

I bought a microsoft band to replace my aging body media fit and wearing both the band reports between 300 and 500 calories a day less than the body media fit. Based on my results so far I'm leaning towards the band being more accurate but I'm not yet sure

 

[Darin Hamel]

I accidentally left mine in exercise mode in the car and had the same super-high heart rates. Crazy. They'll work the bugs out.

 

[Richard Servello]

Depending on your body weight, the incline set, and your intensity, 500 cal in 30 mins is not at all unusual. I've done a 30 min tabata and burned close to 600 calories.

 

[kehoz]

I've used both and would say that, unless you are referring to the Activity Monitor numbers for the Body Media "Core", (which have been messed up for the past year), the other Body Media devices are probably more accurate than the Band.

I also find that the Band numbers tend to be low compared to Body Media (Unless you use exercise mode, where apparently a random number generator is then used . )

Generally, it seems like walking and low-impact activities are not properly accounted for. (For example: I walked over 7.5 miles today and was given about 150 kcal above my normal resting energy... normally would expect around 500-600).

I'm hoping they eventually get the algorithms right. The band has all the same sensors (plus heart rate) as they do, but I don't think Microsoft has done the same calibration and testing that Body Media has done.

 

[dbregman]

My band seemed to account for non workout activities well (I.e. Tennis), but since the latest update -not so much. I played doubles tennis for 3 1/2 hours this morning and my total burn for the day at noon was under 1000 calories..

 

[Skippy07]

My band seemed to account for non workout activities well (I.e. Tennis), but since the latest update -not so much. I played doubles tennis for 3 1/2 hours this morning and my total burn for the day at noon was under 1000 calories..

Interesting, I find that the band significantly over reports calorie burn when I play tennis. For example, the last time I played doubles, I played for 2 hours and 23 minutes with a calorie burn of 1960 according to the Band. I often get somewhere between 2000 and 2200 calories burned when playing doubles at a similar duration (around 2 and a half hours). This seems to be almost double what realistically my calorie burn should be.

-Skippy

 

[teemulehtinen]

Frankly, this is really odd. This weekend, I did a 8km ski tour in the mountains. I skinned up 4km with 500m vertical and then skied it down. Total time used 2h40 and calories burnt just under 1,400 which seem very low to me as it was physically the most exhausting exercise I have done ever since I have had the Band. The total burn for that day 3,600 or so calories and that sn't particularly high either.

Before the update of the band, I would get to 3,300-3,600 range per day just by deskwork and a short 45 minute, intensive weight training exercise. Something is wrong. Or was before. I would like to understand which one - or perhaps both?

 

[dbregman]

Interesting, I find that the band significantly over reports calorie burn when I play tennis. For example, the last time I played doubles, I played for 2 hours and 23 minutes with a calorie burn of 1960 according to the Band. I often get somewhere between 2000 and 2200 calories burned when playing doubles at a similar duration (around 2 and a half hours). This seems to be almost double what realistically my calorie burn should be.

-Skippy

Do you have it in workout mode? In workout mode I think it over reports. In regular mode I think it under reports. Although I am trying an experiment...
When you click on the button, in addition to bringing up the display, it takes a HR reading (I think it used to do this on inbound notificationsas well). I am going to try to click the band a few times during tennis when my heart rate is up. So far on an initial test doing so brought the calorie burn from 120/hour (morning tennis) to 180/hour ( afternoon tennis). I think I only clicked it once.

I will try more extensive testing and report. If they took away the HR check on notification to save battery. That could explain the changes I am seeing.