Need help figuring out your TDEE? Get a Fitbit.

QuietBloom

I have a fitbit that I used to have connected here. I found the synchronization between fitbit and MFP to be unreliable. Some days it worked as it should but some days it not. Is there a trick that I missed?

You can go to your FitBit adjustment (under exercise) and reload the browser window. That forces a sync. I learned this by accident yesterday.

jonnythan

So i'm sure at some point this was asked, I just don't have the time or energy to scan all 11 pages of comments. What exactly does the fitbit track? Is it a high tech pedometer? I guess i'm not understanding how a thing clipped to your waist can tell how long/how well you have slept, or how many calories you burned doing high impact exercise (other than by tracking steps)

Yes, it's a very advanced pedometer. It tracks movement. It knows when you're in a car, because you can't run that fast. It knows when you're sleeping because of the time and no movement. It knows that you should be sleeping, but you're not, because you're moving around. Etc

This is only partially accurate. One of the functions of the Fitbit is a pedometer - that is, it tells you how many steps you've taken. But saying that the Fitbit "is a pedometer" is like saying an iPhone is an MP3 player.

WalkingAlong

So i'm sure at some point this was asked, I just don't have the time or energy to scan all 11 pages of comments. What exactly does the fitbit track? Is it a high tech pedometer? I guess i'm not understanding how a thing clipped to your waist can tell how long/how well you have slept, or how many calories you burned doing high impact exercise (other than by tracking steps)

Well the real answer is that no one quite knows. They don't make the algorithm public.

The device has a 3-axis accelerometer in it. This means it can measure motion, to a pretty high degree of precision, in 3 dimensions. The software translates this motion, via said unknown algorithm, into energy expenditure. It clearly tracks steps, and it can also determine speed to some extent.

There's nothing all that mysterious about burning calories. It takes a certain amount of energy to accelerate a human body of a given mass in a given direction. The Fitbit tracks this and somehow calculates calorie consumption from it. It seems to do it surprisingly well.

I believe it works something like this. There are databases that give the MET values of tons of activities, as measured in calorometry tests on people doing those activities.

MET (Metabolic Equivalent): The ratio of the work metabolic rate to the resting metabolic rate. One MET is defined as 1 kcal/kg/hour and is roughly equivalent to the energy cost of sitting quietly.
http://en.wikipedia.org/wiki/Metabolic_equivalent

So we know that 'walking, 2.5mph, level ground' has a MET value of 2.9. If my BMR is say 1 calorie/min., that implies I'm burning around 1 x 2.9, or 2.9 calories a minute at that activity.

I think at Fitbit they put their device on people in their labs doing all the usual activities it'd track: walking at various speeds, running at various speeds, housework, etc. They correlated the motion info collected by the accelerometer with the activities. Then it's just a matter of hooking it all together. My Fitbit detects the torso (or wrist) motion of a woman walking 2.5mph and applies that MET value to my BMR value and so on for every minute of the day. It picks the closest activity to your motion.

That's why I think it's silly to wear your Flex on your ankle and stuff like that. Whatever motion it's detecting, it's saying 'This is closest to the wrist motion of what everyday activity?'

HRMs estimate calories in a similar way, I believe. They compare your heart rate to tables of average burns at that HR, and apply it to your BMR info. That's why it's wrong to say any of these "measure" your calorie burn or give an "actual" calorie burn. It's going to be accurate for you if you're average.

Leisalynn84

Buying one today so I'm bumping so I can read all of this useful info later : )

Leisalynn84

This makes me sad because I asked for a fitbit one for christmas and didn't get one. I've wanted it for so long, I'm going to have to save up myself. I do a lot of walking for work and it would be insanely helpful for those days.

I foud a zip on craigslist today for $20! Just start looking, I'm sure you can find an affordable one soon.

Charlottesometimes23

Quick question. Can you wear it on a flight without it stuffing up the plane's electronics and causing a crash. Ok, I'm half joking but they always tell you to turn off electronic devices during takeoff and landing.

I'm flying in about 6 hours.

I hate flying.

jonnythan

Quick question. Can you wear it on a flight without it stuffing up the plane's electronics and causing a crash. Ok, I'm half joking but they always tell you to turn off electronic devices during takeoff and landing.

I'm flying in about 6 hours.

I hate flying.

They actually don't anymore. You can use portable electronics like iPads and so forth during takeoff and landing now.

heybales

The device has a 3-axis accelerometer in it. This means it can measure motion, to a pretty high degree of precision, in 3 dimensions. The software translates this motion, via said unknown algorithm, into energy expenditure. It clearly tracks steps, and it can also determine speed to some extent.

There's nothing all that mysterious about burning calories. It takes a certain amount of energy to accelerate a human body of a given mass in a given direction. The Fitbit tracks this and somehow calculates calorie consumption from it. It seems to do it surprisingly well.

Yep, if steps is decently accurate, they have a stride length you can manually correct with tested value, or calculated from your height.
The amount of impact dictates walking or running.
The amount of time between impacts dictates potential length of stride variance compared to that average stride length calc'd, and if an incline is being done based on expected impact (less uphill, more downhill).
From weight and pace and potential incline, then some of the most accurate formula's based on treadmill tests can be used.

I've seen the numbers of some doing treadmill tests at set speed compared to formula calculated, and they almost nailed the formula results exactly.

The stairs feature uses different formula, decently accurate too, but as some have pointed out, seems like accuracy on seeing the stairs taken and the height done is off some times, so that may not be as accurate, but still better than using walking formula.

Non-moving time is based on their calculated BMR, gender, age, weight, height. That's why their initial guess at TDEE is close to MFP, both without exercise expected usually.

That could be one kicker though, awake non-moving would be higher, RMR, than sleeping, BMR. If a great part of your day is non-moving, that part of the day could be a decent underestimate, depending on how big your BMR/RMR difference is.

heybales

HRMs estimate calories in a similar way, I believe. They compare your heart rate to tables of average burns at that HR, and apply it to your BMR info. That's why it's wrong to say any of these "measure" your calorie burn or give an "actual" calorie burn. It's going to be accurate for you if you're average.

Even worse than that. In the sense of potential accuracy, and number of assumed stats.

If there is no VO2max stat self-tested for or manually entered, that must be estimated.
Polar for instance uses probably a slightly tweaked formula from a study that takes BMI and age and gender and estimates a VO2max figure (so assumes worse BMI is worse VO2max).
Their self-test is from another study that includes BMI and resting HR and user selected exercise level, which actually does increase accuracy of calculation. Say 47 ml/kg/min.

Then use assumed 220 - age to calculate a HRmax figure if you don't know your own or better estimate. Say 174.

Then assume the average person has a Lactate Threshold at 80-85% of HRmax. Say 150.

Studies have shown the exercise aerobic level starts at 90 bpm on average.
The Lactate threshold is the other end of that line for the aerobic range, which studies have shown is a straight line formula, before going anaerobic and there being no relation between HR and oxygen use and calorie burn. Say 90 - 150, or 52-86% of HRmax.
There is a studied relationship between VO2max and HRmax, so now you have %VO2max for that range too that corresponds to HR. So now aerobic range is 23%-76% VO2max.

1 MET in studies is usually about VO2 of 3.5 mL/kg/min, which is roughly about 1 cal/kg/hr, so those assumptions used in the MET math.

So now just figure out what % of your VO2 your HR was at, and back track to calories.

Say top of aerobic range, HR 150, or 76% of VO2max of 47 = 35.72 VO2 / 3.5 = 10.2 METS.

10.2 METS x 100 kg = 1021 calories in 1 hr at that level.

Or same math

METS x 3.5 x kg / 200 = calories per min.

Or skip the METS

VO2 x kg / 200 = calories per min.

So calculated or assumed VO2max and HRmax, assumed MET value, all can lead to some interesting results.

Indeed - no where near a measured value. Purely calculated, with a lot of potential wrong assumptions in there.

WalkingAlong

Yep, if steps is decently accurate, they have a stride length you can manually correct with tested value, or calculated from your height.
The amount of impact dictates walking or running.
The amount of time between impacts dictates potential length of stride variance compared to that average stride length calc'd, and if an incline is being done based on expected impact (less uphill, more downhill).
From weight and pace and potential incline, then some of the most accurate formula's based on treadmill tests can be used.

I've seen the numbers of some doing treadmill tests at set speed compared to formula calculated, and they almost nailed the formula results exactly.

The stairs feature uses different formula, decently accurate too, but as some have pointed out, seems like accuracy on seeing the stairs taken and the height done is off some times, so that may not be as accurate, but still better than using walking formula.

Non-moving time is based on their calculated BMR, gender, age, weight, height. That's why their initial guess at TDEE is close to MFP, both without exercise expected usually.

That could be one kicker though, awake non-moving would be higher, RMR, than sleeping, BMR. If a great part of your day is non-moving, that part of the day could be a decent underestimate, depending on how big your BMR/RMR difference is.

The stride length setting for the Fitbit only matters to the distance estimate, not the calorie estimate. And it has not been programmed to detect or adjust for incline.

The BMR they use is really RMR. They apply a MET of .9 to time assumed spent sleeping, like that Wikipedia article I linked to discusses.

The 'stairs' feature is just an altimeter that counts elevation increases of 10' paired with concurrent stepping. It senses barometric pressure.

heybales

The stride length setting for the Fitbit only matters to the distance estimate, not the calorie estimate. And it has not been programmed to detect or adjust for incline.

The BMR they use is really RMR. They apply a MET of .9 to time assumed spent sleeping, like that Wikipedia article I linked to discusses.

The 'stairs' feature is just an altimeter that counts elevation increases of 10' paired with concurrent stepping. It senses barometric pressure.

But distance and then time do give pace, along with using weight, you get the calorie estimate.
Unless there is some FAQ that is specific and they say they don't use it.

I'm going off 11 folks that did testing for me using HRM's and FitBit's and treadmill. The fact their increased calorie burn walking at incline was correctly calculated, tells me it must be figuring out incline somehow even if not that accurately, because these folks also matched up when walking level. Now if I remember right it was a steep 8-10 % incline, and walking 3-3.5, not running I recall.

They must have upgraded their math too, because it used to be you could tag some night time hrs as an event and get a cal / min burn, tag some daytime sitting hrs like at movie, and get the same cal / min burn. The option of telling the FitBit this is sleep time didn't make a difference.
And before they switched from the Mifflin BMR to the WHO study formula for BMR, that calorie burn matched per min.

Do they have a FAQ that says they switched up methods of measuring non-moving time, like their FAQ that says they switched BMR formulas?

WalkingAlong

From Jerrod A., on staff: "Stride length only affects the distance estimation."
http://www.fitbit.com/forums/post/4KTZ2J9SL5T5H/4KTZ2J9SKWN3R/stride length affect calorie

Also from him: "It does not (track incline from a treadmill)."
http://www.fitbit.com/forums/post/4KTZ2J9SLM28F/4KTZ2J9SKWN3R/incline on treadmill

I would think one's torso would move more walking at 10% incline than none. That might explain your test results. While it doesn't notice incline, it might notice that you are moving more robustly?

You might be right about the sleep, it does seem to treat my sleep burn and resting burn the same, though I don't set it to sleep mode so that might change things. I'll try to remember to put it in sleep mode tonight and see. If it's dependent on that then for the user tagging the time as some non-sleep event, it would make sense that it doesn't differentiate whether the low-activity event is in the night or daytime but would assign them the same waking RMR type burn.

GadgetGuy2

I'll try to remember to put it in sleep mode tonight and see. If it's dependent on that then for the user tagging the time as some non-sleep event, it would make sense that it doesn't differentiate whether the low-activity event is in the night or daytime but would assign them the same waking RMR type burn.

You can key sleep intervals retro-actively. The FB website converts the measurements over the time frame into a sleep record. I also seem to remember a profile setting for allowing "naps", I think.

P.S.-Really enjoying this "in the weeds" detail, but then I've always been accurately accused of being anal.

WalkingAlong

I'll try to remember to put it in sleep mode tonight and see. If it's dependent on that then for the user tagging the time as some non-sleep event, it would make sense that it doesn't differentiate whether the low-activity event is in the night or daytime but would assign them the same waking RMR type burn.

You can key sleep intervals retro-actively. The FB website converts the measurements during the night into a sleep record.

P.S.-Really enjoying this "in the weeds" detail, but then I've always been accurately accused of being anal.

Ha! I was just trying that. It didn't change my sleep burn.

I guess it makes sense that it's applied equally across the day. Mifflin is an RMR estimation for a 24 hour period, right? Not much use in trying to allocate a tiny bit less to sleep vs. non-sleep when the totals come out the same. Unless they're going to try to improve upon RMR estimates by tweaking it based on how much you sleep compared to the average assumed in the estimator, but that sounds like too much manipulation for a small benefit.

luckydays27

This thread was probably the single greatest thread I read when I first joined MFP. I went out and immediately bough the fitbit and have not looked back since. I fact I can look all the way around now including to my toes because my belly is smaller after the 38 lb weight loss since I bought my fitbit 8 months ago.

And this has been the EASIEST 38 lb loss I have ever had. I eat an average of 1850 cals everyday and have consistently lost what my goal loss was (started at 1 lb a week loss but now am at .5 lbs a week loss).

Jonnythan is spot on when he says that the fitbit is accurate and can help you. You have to be accurate with your calories in and trust in the science/math/technology of the calories out but it really does work. I, and many others, are living proof of it.

heybales

I'll try to remember to put it in sleep mode tonight and see. If it's dependent on that then for the user tagging the time as some non-sleep event, it would make sense that it doesn't differentiate whether the low-activity event is in the night or daytime but would assign them the same waking RMR type burn.

You can key sleep intervals retro-actively. The FB website converts the measurements during the night into a sleep record.

P.S.-Really enjoying this "in the weeds" detail, but then I've always been accurately accused of being anal.

Ha! I was just trying that. It didn't change my sleep burn.

I guess it makes sense that it's applied equally across the day. Mifflin is an RMR estimation for a 24 hour period, right? Not much use in trying to allocate a tiny bit less to sleep vs. non-sleep when the totals come out the same. Unless they're going to try to improve upon RMR estimates by tweaking it based on how much you sleep compared to the average assumed in the estimator, but that sounds like too much manipulation for a small benefit.

Well, Mifflin and Harris and Katch are considered BMR, test subjects measured for calorie burn upon waking from sleeping at the lab.

Cunningham and Nelson are considered RMR calc's, subjects had been awake, just no food or exercise 5-6 hrs prior to test. Actually, from the ones that comment, they still did it in the morning, but they came in to the lab, so awake and moving already.

The difference between BMR and RMR is generally 150-250 daily, so a mere 8 hr block of sleeping being right, and perhaps 8 hr daytime sedentary job being slightly underestimated isn't much at all for a difference.

I haven't found a free online resource for the WHO formula they switched to, but it seems to be close to Mifflin, depending on degree of being overweight. So a more international study as opposed to strictly American studies, with wider array of participants.

Tendlr

Excellent Read - Bump

dauvis

I have a fitbit that I used to have connected here. I found the synchronization between fitbit and MFP to be unreliable. Some days it worked as it should but some days it not. Is there a trick that I missed?

You can go to your FitBit adjustment (under exercise) and reload the browser window. That forces a sync. I learned this by accident yesterday.

Thanks. I cannot get this to work consistently. I'll have to play around with it to see if I can see a pattern.

TahoeSki

Bump

Ps- I love my Fitbit One! I've worn it every single day since April 2013