Welcome to Debate Club! Please be aware that this is a space for respectful debate, and that your ideas will be challenged here. Please remember to critique the argument, not the author.

Do HRMs on activity trackers really increase accuracy of CO estimates?

moe0303
moe0303 Posts: 934 Member
edited January 2022 in Debate Club
I have read (but I don't have any links at the moment) that heart rate monitors are not a good tool to determine energy expenditure for people. My question is whether or not they are any better than activity trackers which do not have HRMs.

I am very skeptical about HRMs' accuracy in general.
«1

Replies

  • ndj1979
    ndj1979 Posts: 29,139 Member
    they are good for cardio output but not for any kind of weight training...at least that is my understanding.
  • Christine_72
    Christine_72 Posts: 16,049 Member
    Personally I don't trust them, especially not for the exercise I do, which is plain walking... I have a fitbit zip at the moment, and am upgrading to the Fitbit Alta when it comes out, neither have a HR monitor.

    And yes, they are pretty useless for weight training.
  • senecarr
    senecarr Posts: 5,377 Member
    The trackers that do all day HRM don't simply assume that all heart rate changes represent intentional cardio. Most use an algorithm called FirstBeat that attempts to separate the noise from daily activity.
  • EvgeniZyntx
    EvgeniZyntx Posts: 24,208 Member
    edited March 2016
    senecarr wrote: »
    The trackers that do all day HRM don't simply assume that all heart rate changes represent intentional cardio. Most use an algorithm called FirstBeat that attempts to separate the noise from daily activity.

    This might be relevant:

    Firstbeat Technology claims analytic capability in non-steady state. Other companies are able to graph power curves from cycling vs instantaneous HR change. So in reality the old "only for SS cardio" no longer holds for the leaders - Garmin, Suunto, Polar, etc.

    However - HR to nonSS cardio without lab calibration has an error level of about 7-10%. So if your nonSS reading says 200 cals burned it may be 180 to 220. Not too bad.

    Firstbeat does a good job publishing their science:
    https://www.firstbeat.com/science-and-physiology/

    In terms of their utility, you can also use it to estimate things like return to resting, acute exercise stress, etc...
    Definitely useful for the athlete - particularly when trying to figure out long term training response.

    Tools used along with a HRM include TrainingPeaks or GoldCheetah or other analytics tools.

    THANK YOU!
    Someone on here scolded me the other day for suggesting a walker use a heart-rate monitor because it's 'only useful for runners and cycling', but I've used ours (Garmin HRM Tri) for baseline calorie burn on both my walks and my swims, and for walking it came out pretty close to the one my doctor used when he treadmill tested my fitness level. I left the thread, because I didn't feel like arguing. I also think the scolder didn't understand that I don't 'stroll' on my walks; I power-walk.

    Other research also shows that these types of monitors are also more accurate when you also sometimes use them at higher intensity activity. To get an accurate reading at lower HR workouts they need higher intensity data every once in a while. This higher intensity data is used to estimate anaerobic threshold and VO2Max.

    https://www.ncbi.nlm.nih.gov/pubmed/25245124

  • janejellyroll
    janejellyroll Posts: 25,763 Member
    Personal anecdotes aren't worth much overall, but they're all I have to offer this conversation. I eat back the activity adjustments from my Charge HR and have been doing so since July. It's working well. It seems to do a good job, in my case anyway, of separating the "noise" from actual activity.
  • senecarr
    senecarr Posts: 5,377 Member
    senecarr wrote: »
    The trackers that do all day HRM don't simply assume that all heart rate changes represent intentional cardio. Most use an algorithm called FirstBeat that attempts to separate the noise from daily activity.

    This might be relevant:

    Firstbeat Technology claims analytic capability in non-steady state. Other companies are able to graph power curves from cycling vs instantaneous HR change. So in reality the old "only for SS cardio" no longer holds for the leaders - Garmin, Suunto, Polar, etc.

    However - HR to nonSS cardio without lab calibration has an error level of about 7-10%. So if your nonSS reading says 200 cals burned it may be 180 to 220. Not too bad.

    Firstbeat does a good job publishing their science:
    https://www.firstbeat.com/science-and-physiology/

    In terms of their utility, you can also use it to estimate things like return to resting, acute exercise stress, etc...
    Definitely useful for the athlete - particularly when trying to figure out long term training response.

    Tools used along with a HRM include TrainingPeaks or GoldCheetah or other analytics tools.

    THANK YOU!
    Someone on here scolded me the other day for suggesting a walker use a heart-rate monitor because it's 'only useful for runners and cycling', but I've used ours (Garmin HRM Tri) for baseline calorie burn on both my walks and my swims, and for walking it came out pretty close to the one my doctor used when he treadmill tested my fitness level. I left the thread, because I didn't feel like arguing. I also think the scolder didn't understand that I don't 'stroll' on my walks; I power-walk.

    Other research also shows that these types of monitors are also more accurate when you also sometimes use them at higher intensity activity. To get an accurate reading at lower HR workouts they need higher intensity data every once in a while. This higher intensity data is used to estimate anaerobic threshold and VO2Max.

    https://www.ncbi.nlm.nih.gov/pubmed/25245124

    So you encourage all day HRM tracker users to push it to the limit on occasion? :)
  • robertw486
    robertw486 Posts: 2,379 Member
    I would say that with the proper inputs activity trackers with HRM capability have the potential to be better overall indicators of calories out, but many seemed flawed in their basic algorithms and fitness measures to allow that.

    Really without any absolute power measures, HR alone isn't a very good "speedometer". And at times, HR can be fooled somewhat with lag, diet, heat, etc. I know @Azdak recently posted a thread showing how his HRM device could be skewed by other factors vs the machine calculations.

    Though technology and all the understandings are improving, at the current time I don't put a lot of weight on HRMs that don't include other measures of power and/or fitness, with those inputs being critical to the calculations. I can only imagine how crazy my elliptical session calculations might be, having seen my heart rate variances at certain loads and how it sometimes is absolutely NOT a good "speedometer" in the sense of the metered power.
  • hazleyes81
    hazleyes81 Posts: 296 Member
    I have stopped bothering with my Garmin HRM if all I am doing is walking; the calorie reading I get is so close to the Fitbit estimates I don't feel the need.I do use it for my real training, running and biking, partly because I use my heart rate to guide me. The burn for those, especially biking, is usually much different than mfp or Fitbit. Lower for running, higher for biking. I don't use it for strength training. In any case, I do keep my CI from MFP and CO from Fitbit on a spreadsheet and my actual loss is right in line with the estimated based on these numbers, so I do find using an activity tracker and correcting with the HRM to be helpful.
  • auddii
    auddii Posts: 15,357 Member
    I'm sure @heybales has something to say about this too.
  • heybales
    heybales Posts: 18,842 Member
    I'd suggest that basically the premise is wrong, at least for Fitbit and the couple of Garmin's whose data I was allowed to look at.

    The HR-based calorie burn is only used when it's high enough, and usually along with steps going higher showing it's because of movement. Some with Fitbits have even reported turning on Activity timer during their hot yoga with inflated HR, and because of no steps it turns back off thinking elevated HR is for non-exercise reason. Smart device in that case.

    The below exercise level HR is still getting step-based calorie burn for rest of the day. Or assigned BMR level burn if not moving.

    And at least for Fitbit - there seems to be a week or two where it's trying to determine where that line is to start using HR-based - probably based on getting some "resting HR" stats - they don't use what would normally be called a first-thing-in-morning resting HR.

    I've also noticed that the improvements to exercise calorie burn estimate - if no medical heart problems - seems to get matched pretty well with the nicer Polar or Garmins that used the FirstBeat, after that break-in period of getting personal stats.

    I've not seen any claim to still use FirstBeat on the all day trackers though - probably too expensive to license it.
    Public studies share what they are basically doing.
    The flexpoint separating non-exercise and exercise is about 90 bpm men and women, average. Adjust that based on resting HR perhaps.

    I'm really convinced that Fitbit is doing what Polar does with inputted stats and Garmin does with observed stats of exercise.

    The studies Polar and Garmin (Firstbeat) both use rely on BMI, exercise frequency, resting HR (Lifetime athlete option), Age/gender, and estimated HRmax.

    Fitbit like Garmin knows exercise frequency (Polar has you select it, or did on like RS300X type models).
    Resting HR is observed, measured by Polar at VO2 test time (better I think unless Fitbit has unseen stat of true restingHR in morning).
    BMI and gender/age obviously known like Polar has.

    Those all give VO2max estimate.

    % of HRmax assumed for anaerobic threshold, and you got your straight line slope of calorie burn to apply your math too - or use the study formula from the Polar funded study for that matter.

    And while it is interesting how well they can start to match up - matching doesn't mean accurate as pointed out by several in topic.
    Just as HRM being different than database or treadmill doesn't mean it's more accurate either.

    If people actually watched the HR instead of the calorie burn, and tied that to a specific level of effort like settings or pace or such - then they'd see how much HR can fluctuate with other variables outside the effort.
    And if HR is that different, calorie burn is that different, for no reason other than say you didn't drink those 2 glasses of water in the afternoon prior to the workout.
  • robertw486
    robertw486 Posts: 2,379 Member
    @heybales

    A lot of solid input there. Although in all fairness if I had a team of code writers and the ability to dictate how to design an HRM, I would see it as difficult to keep simple enough to be attractive to many users while still having the ability to calculate various calorie burns from various exercise types.

    I think the VO2max tests are a great idea to help with training zones, but at the same time lacking in properly estimating for loads and variables such as the hydration issue you mention. I think to really nail down a good calorie estimate both would have to be accounted for better. Being that the HR is the largest display on my elliptical, and I can also watch pace as well as cadence and load (watts or calories per minute) I completely agree with your statements regarding HR variance.

    I'm almost certain I could easily fool most HRMs into inflating my calorie burn vs machine power numbers. At lower levels of intensity it might be harder, but at moderate to upper levels, fairly easy most likely. Being that my chest strap pairs with my machine, two way communication to a machine that has a true power measure would probably me the most accurate thing, but take the calculation out of the hands of the HRM designers somewhat. I would think some type of two way communication could vastly help the HRM by giving some measure of absolute power, be it measured in watts, or calculated from mass x velocity such as could easily be done with a treadmill.


    To be honest at the present time I don't think most HRMs do much for calorie calculations that the proper app doesn't. They do have some features that make things easier, but not really always better or more accurate.
  • mattyc772014
    mattyc772014 Posts: 3,543 Member
    I have been using a Fitbit Surge for over a year now and can say it is fun to use, keeps me engaged with exercise more and seems accurate. But I find myself always going back to my own data to keep myself within calories etc. Basically what I am saying is I could lose my Surge watch tomorrow and not blink an eye at what my calorie and exercise goals are or should be.

    I do also have a Polar chest HRM with my elliptical at home and have used both while exercising. The chest HRM is more accurate with HR than the Surge. I also see that I burn more calories with chest HRM than the Surge at the end of my workout. Why, because it takes about 5 mins into the workout for the chest HRM to sync with the elliptical.

    Aside from HRM, the Surge is most beneficial for tracking a run or walk. I can get my per mile pace as I run and really enjoy that. I like the breakdown of my pace per mile and distance within the Fitbit app.

    My takeaway...don't trust AI. :)
  • heybales
    heybales Posts: 18,842 Member
    @robertw486 -

    I installed some new but old model Shimano cranks finally in expectation of getting Watteam PM - alas they say not compatible with that old one. I'm going to ask if really tested and doesn't work - or merely not tested but won't support.

    Anyway, I wish I had actual watts on bike (though that post-workout analysis utility we talked about prior has seen great success) like you see on the elliptical, is that newer or older unit that has that?

    I've seen the studies looking at mechanical efficiency involving the legs - 22-24% - I've never looked for studies that include body like elliptical, rowing, cross-country skiing, in case the use of upper body lowers the %.

    But that would be so easy to get really decent accurate calorie burn then.
    Have you ever done math on elliptical reported calorie burn and average watts to see what % it might be using?

    Then that wireless protocol from HRM strap to machine could be made 2-way perhaps (not sure how you have HRM watch besides Garmin, only expecting to see heart beats allow something else coming in), and instant watts given back to HRM to do simple math with and perhaps log the info too.

    That would be so cool to download Garmin data and have your power info just like using a bike power meter, but from the rower or elliptical.

    That simple Polar VO2 test they do, I'm sure they realize it's limited for good training, and the reason they don't have options to base training zones on the test's results.

    Ever tried changing your breathing pattern with Garmin (that was your HRM I thought) to see it change estimated calorie burn?
  • EvgeniZyntx
    EvgeniZyntx Posts: 24,208 Member
    senecarr wrote: »
    senecarr wrote: »
    The trackers that do all day HRM don't simply assume that all heart rate changes represent intentional cardio. Most use an algorithm called FirstBeat that attempts to separate the noise from daily activity.

    This might be relevant:

    Firstbeat Technology claims analytic capability in non-steady state. Other companies are able to graph power curves from cycling vs instantaneous HR change. So in reality the old "only for SS cardio" no longer holds for the leaders - Garmin, Suunto, Polar, etc.

    However - HR to nonSS cardio without lab calibration has an error level of about 7-10%. So if your nonSS reading says 200 cals burned it may be 180 to 220. Not too bad.

    Firstbeat does a good job publishing their science:
    https://www.firstbeat.com/science-and-physiology/

    In terms of their utility, you can also use it to estimate things like return to resting, acute exercise stress, etc...
    Definitely useful for the athlete - particularly when trying to figure out long term training response.

    Tools used along with a HRM include TrainingPeaks or GoldCheetah or other analytics tools.

    THANK YOU!
    Someone on here scolded me the other day for suggesting a walker use a heart-rate monitor because it's 'only useful for runners and cycling', but I've used ours (Garmin HRM Tri) for baseline calorie burn on both my walks and my swims, and for walking it came out pretty close to the one my doctor used when he treadmill tested my fitness level. I left the thread, because I didn't feel like arguing. I also think the scolder didn't understand that I don't 'stroll' on my walks; I power-walk.

    Other research also shows that these types of monitors are also more accurate when you also sometimes use them at higher intensity activity. To get an accurate reading at lower HR workouts they need higher intensity data every once in a while. This higher intensity data is used to estimate anaerobic threshold and VO2Max.

    https://www.ncbi.nlm.nih.gov/pubmed/25245124

    So you encourage all day HRM tracker users to push it to the limit on occasion? :)

    I encourage everyone to push it to the limit on occasion. Literally and figuratively. (with the support of your doctor, yada yada.)

  • robertw486
    robertw486 Posts: 2,379 Member
    @heybales

    I wish I had meters for the bike, but really at the end of the day can't justify the expense. I just use Strava and accept that error will exist. Was it Training Peaks that you were speaking of? I have several of the bike discussions bookmarked, but that came to mind, as well as Bikecalculator.

    I haven't done all the math on the elliptical, but due to the readouts would have to "reverse engineer". The machine is a Precor and for years most of their machines have used an eddy current resistance system. This is essentially a generator that produces a magnetic field and applies resistance. It's the same design as a Velotron bike ergometer. But due to limitations on the readouts, the one single metric that is most useful in final calculations is the calorie burn. Though it displays wattage, strides per minute, METs, and calories per minute, they are all instantaneous readings without any average at the end of the workout. In the case of watts, it is wattage per stroke, not the higher reading that would be true power level. Since watts readout is whole numbers, doing the math could have a decent bit of error. Half a watt at 150 strides per minute adds up quick.
    So finding the efficiency they used is really virtually impossible, but I could probably get close. I can use total calories burn to calculate wattage, as well as calculate the net vs gross, since the machine readout gives gross calorie burn based in input weight of the user.


    But here is my thinking for the best integration of HRMs to devices that may or may not be available.

    1. True power measures if available, such as meters on a bike or wattage on my elliptical.
    2. Power measure estimates from mass x velocity, such as true run pace on the flat or treadmill
    3. Power measure estimates as above, but with input for elevation

    With any input beyond HR, the HRM computer would have the ability to know or at least calculate net power. On the devices with meters, it would be known. But on the mass x velocity calculations, it could at least use absolute energy requirements.

    With full two way communication on some devices, the ability to chart and better understand load or power vs HR would also allow better learning ability for the HRM device. This should be helpful in possibly understanding efficiency on any given workout. I would also think it would be helpful in estimations where GPS or distance measures are not known, since it could "learn" the ramp up (and down) rates of HR vs power. Combined with possible cadence information from steps or movement, even better. **As a note on the elliptical similar to a bike I can always put out more power at higher cadences, and as such power measures alone don't tell the whole story IMO.** The VO2max testing methods in place seem OK for the present, but other than for training zones I really see it as a useless metric. I would think true VO2 uptake for any given workout would be a better metric and a smart HRM would be able to calculate that fairly close.

    At this time I have a chest strap that pairs with both my machine and a (cheap and simple) biking computer. I use machine readouts for my calorie burns, and don't have a HRM type device to calculate. But points taken on the breathing patterns as well. As someone who has done a lot of diving, snorkeling, etc, I'm really good at blowing off carbon dioxide to bring my HR down, usually 4-5 BPM even at higher outputs.


    I think integration with machines would be a great thing to allow the HRM to learn. @Azdak recently posted a machine vs HRM thread to show where flaws exist. The trick IMO would be figuring out where flaws in either the HRM or the machine exist to help the HRM learn better.



    I'm also interested in the comments by @mattyc772014 and @rainbowbow (in another thread) about HRM pairing to machines. I'm not sure if any actual devices pair or not, or if so what information is shared. I'm wondering if it's just the chest straps pairing, or anything other than HR is shared. But I guess it's possible that two way communications are already being used.
  • EvgeniZyntx
    EvgeniZyntx Posts: 24,208 Member
    How do you overcome the efficiency conversion estimate between power and metabolic calories? That is certainly neither constant nor well-calculated.
  • robertw486
    robertw486 Posts: 2,379 Member
    How do you overcome the efficiency conversion estimate between power and metabolic calories? That is certainly neither constant nor well-calculated.

    I think that would be the hardest one to try to determine. Short of lab gas exchange testing it would all be just a guess based on absolute VO2 vs power estimations. But I would think that if the HRM device were smart enough to learn from known absolute power measures, you could at least determine (as a possible example) that efficiency on the bike was higher or lower than efficiency running. Being that most studies seem to mostly agree on O2 use efficiency, that could be a baseline for the models.

    As an example, I've seen studies where the O2 requirement for running vs biking at a setpoint wattage, the running shows higher O2 demand. This exposes a "flaw" where net power vs gross power is somewhat skewed, and that even a power measure we can assume absolute is not exacting.

    I would think that part of any estimations would be a slippery slope either way, but there may be sufficient information short of the lab testing to go with averages per any given fitness level, age, etc.
  • mattyc772014
    mattyc772014 Posts: 3,543 Member
    @robertw486 My elliptical came with a non coded Polar chest HRM that syncs with it. I dont think I can sync a different device with my elliptical unless its non coded. Technically I could use the chest strap with another machine lets say at the gym, but because it is non coded I will get interference from the other machines. Must be manufacturer programmed on a 2Ghz frequency which I think are the majority of electronics. Hence the delayed syncing I would imagine.
  • ParadiseLost91
    ParadiseLost91 Posts: 28 Member
    Who cares if it's 20 calories off once in a while? I wear my FitBit Charge HR daily and find it a big help in my weight loss.
    I eat my exercise calories (measured by FitBit) because I'm at net 1200 calories, and I am not gaining weight, actually losing it quite fast.

    It's a huge motivation to have a fitness tracker for many people, because you can see how well you did, and you can see a difference between workouts where you didn't push yourself, and workouts where you did.
    If you plot one hour of spinning into MFP, it will always give you the same calories. If you use FitBit, you will get more calories the harder you worked.

    So the technology might not be perfect, maybe it's 5 or 20 calories off. Why is that so important? Isn't it better that having a HR wrist band makes you motivated to push yourself?

    As I already said mine doesn't feel inaccurate at all.
This discussion has been closed.