Calibrating Calories Burned Using Weight Loss Data

[RichardCollins]

I posted this article in the Charge HR help area, but I thought it might be of interest to any Fitbit user trying to lose weight.  I used my downloaded data and actual weight loss data to see what multiplier I needed to use to adjust the Fitbit Calories Burned to something appropriate for me.  I knew from another study that the Active Calories are overestimated, so I suspected Calories Burned are too. 

To not keep you in suspense, I have to multiply whatever Fitbit reports for Active Calories by 72% (0.72) to get a more accurate value.  And I have to multiply Calories Burned by 85% (0.85) to get the appropriate Calories Burned that corresponds to my actual weight loss. 

My exercise is walking.  Similar calibration techniques apply to other types of exercise.

Fitbit has been extremely helpful to me in losing weight.  I have lost 25 pounds in just over two months.

Calibrating Calories Burned Using Weight Loss Data

Calibrating Active Calories Using Single Activity Log

[LizziePooh]

Thank you for the information. I just bought the Charge and not the Charge HR not realizing that the charge does not calculate heart rate.

So to clarify, you have the Charge HR and you have to adjust what it shows?

Do you know if the Charge calculates more accurately or does it do the same thing, overestimate calories burned? I was hoping it underestimated. lol!

Thanks for your help and for sharing the formula. If it is the same for charge, I will start using it. I don't want to waste time limiting my eating lol! if I am not getting enough calories burned. 🙂 I might as well stay fat. 😛

[RichardCollins]

@LizziePooh

The Charge and Charge HR use different methods for estimating calories, so I am not sure whether they are over or under-estimating Active Calories and Total Calories for the Charge.  The Charge HR uses some kind of correlation between heart rate and calories burned, but either the implementation of it by Fitbit is off, or they just did a bad job of programming and accounting for calories.  The Charge HR that I have reports calorie estimates must be adjusted.  I have to multiply my Active Calories by 72% and my Total Calories Burned by 85% to reduce them closer to what I actually burn.

The Charge only has an accelerometer, so it is sensing change in velocity.  For footsteps it detects when your foot hits the ground and sends a shock up to your wrist.  That burst of acceleration is read as a step.  They either look up your stride from a table by your height, age and sex, or use what you enter into the Settings page.  They multiply the stride times the steps to get miles.

Miles = Stride * Steps/(5280*12)

I have to guess, but they must use a correlation between integrated (summed) acceleration and calories burned based on some kind of calibration exercise.  "The more you move, the more you burn" kind of approach.

In the Charge HR, Fitbit will give you credit for a faster heart rate, even if you are not taking steps.  I have a friend I met on this forum who was sick.  It raised his heart rate and made it look like he was burning more calories.  Perhaps he was, but he was less active.

I bought my daughter a Charge.  I will ask her to send me some details on one of her activities.  I can check whether their estimate is better or worse.  If you can record one of your walks and tell me the steps, miles, duration in minutes and seconds, and the calories burned, and what distance you walked according to a map, I can tell you what they are doing.  Or you can do it yourself:

Calibrating Active Calories Using Single Activity Log

Calibrating Stride Using Google Maps I

[LizziePooh]

I love how detailed and informative you are, Richard. I was going to have you calulcate it but since you went through and took the time to detail such clear and informative steps, I am going to try it.

Here are the numbers I currently have just for your daughter's comparison. I still need to get my distance from the map but you even included instructions on how to do that. Love that. Once I have done that, I will give you that piece of info.

Have a good day. 🙂

Total Steps: 6335

Miles: 2.55 miles

Duration: 1:01:11 (59 active minutes)

Calories Burned: 409

@RichardCollins wrote:

@LizziePooh

The Charge and Charge HR use different methods for estimating calories, so I am not sure whether they are over or under-estimating Active Calories and Total Calories for the Charge.  The Charge HR uses some kind of correlation between heart rate and calories burned, but either the implementation of it by Fitbit is off, or they just did a bad job of programming and accounting for calories.  The Charge HR that I have reports calorie estimates must be adjusted.  I have to multiply my Active Calories by 72% and my Total Calories Burned by 85% to reduce them closer to what I actually burn.

The Charge only has an accelerometer, so it is sensing change in velocity.  For footsteps it detects when your foot hits the ground and sends a shock up to your wrist.  That burst of acceleration is read as a step.  They either look up your stride from a table by your height, age and sex, or use what you enter into the Settings page.  They multiply the stride times the steps to get miles.

Miles = Stride * Steps/(5280*12)

I have to guess, but they must use a correlation between integrated (summed) acceleration and calories burned based on some kind of calibration exercise.  "The more you move, the more you burn" kind of approach.

In the Charge HR, Fitbit will give you credit for a faster heart rate, even if you are not taking steps.  I have a friend I met on this forum who was sick.  It raised his heart rate and made it look like he was burning more calories.  Perhaps he was, but he was less active.

I bought my daughter a Charge.  I will ask her to send me some details on one of her activities.  I can check whether their estimate is better or worse.  If you can record one of your walks and tell me the steps, miles, duration in minutes and seconds, and the calories burned, and what distance you walked according to a map, I can tell you what they are doing.  Or you can do it yourself:

Calibrating Active Calories Using Single Activity Log

Calibrating Stride Using Google Maps I

---

[LizziePooh]

Hi Richard,

I did the google map measurement and I came up with a differece. Fitbit calculated 2.55 miles and google calculated 3.38 mil. Since it took me 1:00 and I typical walk abt 20 minutes per mile, I think google is more accurate.

I did the math you suggested and I came up with a stride for me of 33.93 vs fitbit's 25.51. Does that sound right, do you think?

Here is the info:

Steps 6335

Miles 2.55 miles

duration: 1:01:11

calories burned 409

Google Map distance: 3.38 miles

If you do think it is right, I will take you up on your offer to calculate what I need to enter in Fitbit for me. I want it to be accurate and trust you more than me on that. Thanks! 🙂

[RichardCollins]

FitbitStride = FitbitMiles * 5280 * 12 / Steps

                  = 2.55 * 5280 * 12 / 6335

                  = 25.04 inches that Fitbit uses

YourStride = GoogleMapMiles * 5280 * 12 / Steps

                  = 3.38 * 5280 * 12/6335

                  = 33.79 inches for your actual stride

                  = 2 feet and 7.79 inches for the Settings page

I am being explicit for anyone reading this later.

Just for people reading, you are walking at this speed:

MilesPerHour = GoogleMapMiles/HoursWalked

                       = 3.38 / (1hour + (1 minute/60) + (1 second/3600))

                      = 3.38 miles/ 1.0169 hours

                       = 3.32 mph

Now go to this ShapeSense Walking Calorie Calculator and enter your weight, GoogleMapMiles, and duration.  Note the Gross Calories they tell you.  Then go to the ShapeSense Net Calorie Calculator to convert those to Net Calories (calories only from activity, not base metabolic rate).  This is the number of real Active Calories you are burning during your walk.  I suspect you will find Fitbit way overestimating it, even if they were way off in your distance.

You will need to go to the Settings page and put in your actual stride.  (You probably have already done that).  Then go for another walk.  Fitbit will make an estimate of the calories for that walk.  Check the Net Calories from ShapeSense.  Calculate your Calorie Multiplier.

CalorieMultiplier = NetCalories/FitbitCalories

This will be your guide to shrinking Fitbit Active Calories to a more reasonable size.  My CM is 72%.  I have to multiply Fitbit Active Calories by 72% to get my Net Calories.

[LizziePooh]

@RichardCollins

Thanks for the info, Richard. I will check this out in the near future.

Have a good day/night. 🙂

[Heybales]

Just FYI - you don't want to use Net calories if thinking of overwriting what Fitbit had as calorie burn.

Why?

Every database, HRM, Fitbit, treadmill, whatever - uses Gross as result. Not NET unless they have option, like a couple of sites do.

But when replacing the Fitbit calorie burn, you do want to include your resting metabolism that NET would exclude - because you did burn that in addition to the calories strictly for the exercise.

All those sites are telling you what you burn in total in that time (estimated of course) - and if replacing what Fitbit came up with - that's the figure you want to use - Gross.

So of course Fitbit will appear to be overestimating - but it's actually not - because you are comparing apples and oranges if comparing NET and GROSS.

Fitbit actually has a better chance of underestimating daily calorie burn.

It assigns sleeping level calorie burn, BMR, to all non-moving time.

But when awake you actually burn more - RMR.

When standing not moving you burn even more.

When eating you burn about 10% of calories eaten to process and digest food.

That is all unaccounted for by Fitbit.

I think there's a better chance your food logging is off enough to cause the difference you saw.

Or the fact you can actually change what your body burns by eating too little, but have no idea if you did that to yourself. But many do. That changes the equation too.

Or as many have done, burned off muscle mass from past yo-yo diets, and so Fitbit's estimate of daily burn is for average body with average muscle mass, not lower.

Of course the method of changing the estimate to match results is just fine, but to suggest that Fitbit is always going to be that far off flies in the face of all the people that lose exactly what the math suggests they would, or more.

Or me, I have to increase the height in Fitbit stats, because I have more muscle mass than average - so to get the higher calorie burn that is correct, Fitbit has to be fooled in to thinking I burn more as BMR, by being taller.

[Sad_Panda]

Stay healthy!  That seems like a lot of weight for just two months.