Fitbit calorie burn very different than iphone

OllieS0806

I have been previously using my iPhone to feed step/calorie data to MFP. For example, I get iOS credit for 2053 calories for 36k steps. That is about 0.08 calories per step.

I just got a fitbit today, and I am getting way more credit. As of now, I have 13k steps, but have received a 2500 calorie credit. That is about 0.18 calories per step, almost double iOS.

How can these be so off? I have been using the iOS for about 4 weeks now with MFP and am comfortable with my rate of weight loss. I don't want to have to manually adjust the exercise credit in my head every day. Is there a way to change the cal/step? Or was iOS way off? For reference, I am a 190lb male.

Thanks

heybales

They are dealt with differently.

Using the iPhone did NOT feed calories to MFP, only steps.
It was up to MFP to do a super rough calc of what that means for calories. It has no idea of distance, so it's estimating a stride length based on height and gender (just like Fitbit does for that setting), which is actually the best estimate of calories, because pace and mass is very accurate calculation.
But that's a static figure, same per step.

But your stride length changes, Fitbit can figure that out from impact seen by accelerometer.

Fitbit is calculating calories for daily activity based on distance too, it knows the pace and dynamically adjusts that distance (grocery store shuffle to exercise walk).
Steps is merely a figure for displaying. Daily calories burned is another figure MFP receives to actually do math with.

I wouldn't even attempt math when you are working with apples and oranges. (ah!, guess which is the orange)

Now, if the Fitbit sees elevated HR and enough steps it could be slipping into exercise mode, and using HR-based calorie burn, which is an inflated calorie burn method for walking.
So you might see if there are some workouts auto-created.

If you took a purposeful walk, and know the start/duration time - you can create an Activity Record on Fitbit (even if there is a workout record there already - in which case use that start/duration/distance shown).

That's just a snapshot of the stats for the stated chunk of time - it's not adding anything, merely letting you view it.

See what Fitbit estimated for calorie burn for the given distance and time, and compare to the Gross option here.
https://exrx.net/Calculators/WalkRunMETs

That compare does NOT work for taking daily distance and time - needs to be a purposeful walk.

That should tell you if Fitbit is close to correct at least.

OllieS0806

Thanks for the reply. I checked one of the logged activities, and it looks like Fitbit is somewhat higher. I didn't know how to translate the 28 feet of elevation into "grade" however. For the exercise I picked, I got 190 calories on EXRX, and 221 on Fitbit.

heybales

28 ft isn't enough too bad for short distance to worry about.

So it sounds like it was a workout on Fitbit, then it used HR-based calorie burn - which is indeed inflated at the lower and upper ends of the aerobic zone. So right above daily (like walking), and right below anaerobic (like sprinting).

That would be a case where logging your own Workout Record (different than Activity Record snapshot of existing stats) with stated distance and start/duration times - and let Fitbit calculate calories from that. It's using pretty much the same well established formula as exrx.

Now in that case, your Workout Record will overwrite the stats in your daily record, in this case calories is all that changed.

You can leave the Activity Record that was created with HR, and distance, calories burn, ect when you selected you were doing a workout. Those stats will be overwritten in the daily stats, of course calories is the only difference.
But it allows you to still view HR info, which is good for review.

Just don't be concerned you have 2 what appears records there, only the latest matters for info. Fitbit is a replace only system.
So long as you get the start time and duration matching, no doubling effect.

So that sort of inaccuracy is likely one aspect of the distance you saw.

It could also be the daily activity steps is mathing out to different distances to calculate calorie burn from.
Ever walked a known distance and confirmed the Fitbit saw it right?
If you can get 1/2 to 1 mile, and the difficult part, about 1.8 mph pace, you can confirm the Fitbit has the correct stride length.

And then it'll dynamically adjust each step distance from slow pace to exercise pace decently, hence the 1.8 being middle of the road.
Because you do have an enormous amount of steps - so distance matters now.

OllieS0806

The distance looks accurate. I have taken a walk with my old Pacer app, and the Fitbit logged the same distance. It has the GPS feature anyway, and the mapped walk looks right. I accidentally logged it as a run, so not sure if that matters in any way.

I also noticed that my heartrate tends to be higher on walks when I have my dog with me, so I assume that is because I have to do some pulling from time to time. The walk I took without my dog, even though much longer, showed almost the entire thing in "fat burn" zone, while my longer walk with a faster pace showed very little "fat burn" zone. Maybe this thing is accurate?

heybales

GPS isn't used for tracking daily distance, inside and whatnot - stride length is.

Recommendation is to test the stride length setting, not test accuracy of GPS, since GPS not used for rest of the daily steps. Trying to recall how to get the step-based distance from a walk that has the GPS distance on it too - it is a view because people complain about the difference wondering why.

HR-based calorie burn is inflated for walking.
What HRzone happens to be in is immaterial to that formula. (I'd suggest that if a walk goes above "fat burn zone" then your HRmax used by Fitbit is wrong and that's a whole other issue for workouts being inflated)
The increased HR and therefore calorie burn from some times of pulling dog would be true, but not the amount HR would indicate as a difference.

So yes the Fitbit is probably more accurate than MFP using steps from iPhone, but you for sure are getting inflated burns too.

OllieS0806

OK thanks. Is there a place in MFP where I can just put in some kind of exercise offset or something? Or do I just have to do the math in my head?

heybales

For walking or running, frankly I'd still start the workout on the Fitbit when you start doing it.
When finished that is an Activity Record, merely showing the stats for that chunk of time, side effect was starting the HR-based calorie burn and per second HR logging.

Afterwards, like just sometime before your next meal if that calorie burn is going to effect it - make your own Workout Record, using the already stated Start, Duration, and Distance given - take the option to left Fitbit calculate the calorie burn.

That record will be used in your daily stats for distance and calorie burn. It'll replace not add to.
The new Daily burn will be sent to MFP to do math with.

Now, tracker accounts are only supposed to send new data when the calorie burn goes up 100 higher than prior sync - so if you just lowered the daily burn with more accurate exercise info - the next sync could take awhile longer depending on how much you lowered it.

My Sat long ride was actually 1000 cal lower than initially synced estimate - it took food updates to bring the new amount over because I'd never have gone up enough as sedentary as I was after the late ride.

OllieS0806

I also reduced my stride length a bit, as I believe my legs are shorter than average even for my height. Should I expect to see a calorie update?

OllieS0806

I added those manual activities like you said. So I don't have to delete the fitbit-generated record?

heybales

That is an Adjustment record on MFP based on that Fitbit Total Daily burned figure - as that figure comes across every increase of 100 cal through the day - the math is updated.
No need to touch it, it'll change.

You can tap and hold on that adjustment to see the details of when the last Fitbit sync occurred and the figure at that time. When you see about 100 more on the Fitbit, know it'll be about syncing time.

After your next sync with device and your Fitbit account, stride length on device will be updated, and the math will result is less distance and calorie burn.
So when that happens going forward you'll get smaller movement based burns.

OllieS0806

Thanks for your help. This is going to take some adjusting. The difference is so extreme. My exercise total on 36k steps the old way was 3053. The new formula has me about 4k calories for 40k steps. I will see how it goes tomorrow with the updated information on the stride.

heybales

Well, like I mentioned that website formula cannot be used for anything outside of a purposeful walk.
It'll be wrong for taking say a block of 8 hrs of a workday with lots of stop and go, and trying to pull an average.

Now if you are saying you are literally going out for a long walk/run of 40k steps at one time - then indeed the formula method is best.

And you can indeed do what it sounds like you've done - revisit some past days and do it there just to see what happens. You've tweaked the stride-length since then, but still an idea.

OllieS0806

I've only overridden the purposeful walks, but that is not my entire day's worth of walking of course. I pace around the house a lot when on conference calls, go to the grocery store, etc.. According to the log, it looks like about half the day is purposeful walks.

The difference still seems extreme, but I'll give it a few days to see how it goes now that I have made the adjustments.

heybales

Just some insight as to the possible effect if many of those steps is what I call grocery store shuffle.

The accelerometers for years have the logic built into them to be given a mass and stride length, and it calculates what the expected impact of a step would be for that.
Depending on the impact actually seen, a calculated distance is spit back out for what that step distance likely was.
It can work surprisingly well considering what's being calculated. (elevation changes throw it off, steps up and down are opposite really)

As you can imagine, actual impact matching expected impact is the stride length, and the most accurate place in the range of what is possible.
The farther the impact is from that expected spot, the less accurate it's going to become.

Like if stride length was set for the long stride of a purposeful exercise walk that was about as fast as you could go, then your grocery store shuffle would be totally opposite that and be at worst possible accuracy.
And depending on where you really got most your steps from, that could be a bad effect for calories.

That's why accuracy of stride length in the middle of your walking pace range, means it can adjust up and down from slow to fast - and hopefully those other extremes are not so far away to get badly accurate.
Usually 1.8 mph is that pace.

That is a hard pace to do for 1/2 mile or more, especially on a track with people watching - it just looks like you are lolly-gagging along.
But starting a workout for that test, getting the steps after, and mathing out the distance and steps - gives a very accurate stride length for middle of the road.

Interestingly, research has shown that changes with weight too, at least at the exercise levels mostly, as it's easier to move. But the calculated stride length everyone uses appears to still use the gender/height formula. (I don't think I saw one that includes weight, because who knows how that would impact someone).

OllieS0806

Things are looking more reasonable (in-line with MFP estimates). One thing odd: when I override an exercise, the step count for the manual entry is much higher (4300 steps vs 3200 for a 1.2 mile walk). Does this mean that my manual stride length is on the low side?

heybales

Good catch.

A manual walk/run entry uses the given distance divided by stride-length to get a step count for it.
(sadly those steps are also invalid for challenges since that's how people used to cheat for high step counts)

It depends on where the inaccuracy was.

Was the 1.2 mile used on the manual entry matching what Fitbit said for the distance?
Or that was measured known distance from other means?

Did Fitbit get that distance from GPS or steps/stride-length?

OllieS0806

The 1.2 miles was from Fitbit. I don't know how it computed the distance though. It does have its own GPS, so either method is possible.

OllieS0806

If I just don't record purposeful walks, can I avoid all of the overriding stuff?

heybales

And confirm the auto-start workout is disabled.
Then you can leave as is.

If Fitbit see HR go up enough with enough steps, it'll auto-start a workout and enabled HR-based calorie burn anyway.

But indeed, if that is disabled - all walks will be done by distance for calorie burn.

If one of them makes you curious later as to the stats on it to compare (like did I do it faster, lower HR, what distance, ect), just manually create an Activity Record (not Workout Record) - and you'll be shown the stats Fitbit saw for that chunk of time.

That may be the only hard part - when did I start it? But usually you can look at your daily graph and tell when it started and how long it must have been.

OllieS0806

I turned off auto-logging of exercise and stopped logging it manually also. However, I think it is now giving me credit for zone minutes that happen when I exercise even without recording it.

On days where I overrode the exercise, I had 33k steps and about 3472 calories. The next day, I stopped recording the exercises. I ended up with 34k steps but 4300 calories.

I am almost done with this device. I don't understand why they cannot fix the algorithms for calorie burn that basically everyone knows are overstating it. Then I have to stop using half the features that it is capable of.

I would rather go back to letting MFP calculate my calories strictly based on step count. Does anyone know if the apple watch integration is better, meaning it only counts steps and lets MFP figure out calories?

heybales

Zone minutes is merely a figure like glasses of water, just something to see, not used in any calcs.
Credit is for the figure, not calories, if you did like you say, not starting a workout where HR matters.

What was the distance on each of the days you are comparing?

Because again - it's distance for calories, not steps. Steps merely gives a rough idea of how active you must be - and 33K is huge.
When you have that many steps, small per step inaccuracy can be rather big.
And differences in distance can be big.

1000 cal difference is rather big too though, but with that many steps I've seen a huge difference in distance too and resulting calorie burn.

Giving MFP only a step count and letting it figure out distance (which it doesn't show) and resulting calorie burn is even rougher estimate.
Lower doesn't mean more accurate when you only have 2 figures to compare.

During that 33K or 34K day - did you have a purposeful exercise walk/run for a big block of time? (not daily stuff interrupted with standing around)
If so, did you confirm calorie burn by creating an Activity Record (where you provide start/end time only) and seeing the stats for distance and calorie burn, and then compare that with the online formula for calorie burn.
Might be surprised.

Apple watch and Fitbit can both be set to only get steps from the device - must disconnect accounts for Fitbit to be reused that way.

OllieS0806

I did about the same amount of purposeful walk/exercise on both days.

So how do I use Fitbit only for step count? I disconnected it, but I don't know how to make that work.

heybales

Once disconnect on both sides - the accounts won't talk to each other.

Then in MFP app - you go to setting for Step Source and select Fitbit.
That means just get a step count reading from Device directly.

OllieS0806

Thanks for all of the help, but I gave up on the Fitbit. Too much configuration to try to get the calories in-line. I am surprised Fitbit continues using the formulas they do when it appears they are known to overestimate. All of the overriding and manual adjustments still resulted in incorrect calories and was a lot of work.

I ended up with an Apple Watch SE, which seems to be much more in-line with expectations. Not sure how the formula actually works, but the extra calories I earn seem much lower and sensible.

heybales

Fitbit does have a few models where if you start a Weight Lifting workout, it doesn't log calories per HR, but rather per database rate of burn.

They would be smart to keep using distance-based calorie burn for workouts start as Walking or Running instead of HR, or like Garmin run both calc's and weight the distance for about 80% of the calorie burn, 20% based on the HR.

For the other they are going for average usage - where these inaccuracies may cause upwards of 5% off either direction for vast majority of users.

You aren't the average user likely.

I know of 3 and seen their math that shows they are underestimated by about 15% - they'd be losing weight following the numbers for maintenance.

PAV8888

Hey guys: in all this discussion have you taken into account that (unless I am mistaken) Apple is reporting active minute calories but fitbit is reporting total calories i.e. it is including the calories you burn by being alive?

The MFP integration (which works for Fitbit but not for Apple) then takes care of these being alive calories.

So the fact that you're seeing 4000 Cal for an activity on Fitbit does not mean that you are actually getting an extra 4000 Cal added to MFP.

heybales

I was mainly having OP look at why the Fitbit was higher daily calorie burn, several things could be part of issue.
They get a lot of steps, so distance accuracy matters for the daily stuff.
The exercise is stuff that HR-based calorie burns will be inflated, always requiring a manual workout entry to get better calorie burn.
Just too many annoyances.

Apple actually reports just one daily burn figure to MFP when accounts are synced - their base or sedentary calorie burn, pretty close to MFP Sedentary burn rate.

They don't report the active calories actually.
They do send workouts.
Which makes what they send a whole lot worse with the math.