After Update

[gadgetrants]

I would think that MS uses a standard calculation for that.

I'm afraid we may have to part ways at that assumption -- I'd like to delude myself that the MS engineers are more clever, and since they have a good deal more biometric data (e.g., the GSR sensor) at their disposal, the caloric estimates are more fine-tuned than what the generic equations can do. There are a few factors supporting my delusion (or naive hopefulness), but I'll sketch here one or two of them:

My sense of calorie-burn estimating formulas is that they take two major inputs: (1) relatively personalized data that characterizes you by using, for example, your height, weight, age, sex, etc., and (2) the duration and "intensity" of activity you're performing, for example, running, swimming, biking, etc. It seems that for (2) many approaches use a generic activity "multiplier," which, like lumping you in with a class of other people who have comparable physical characteristics, is a rough heuristic at best.

But there's another approach. That is to measure, say, 6 or 7 things that can be recorded on the Band, including acceleration, HR, skin temp, etc. while having a bunch of (say, 100) people actually do stuff like run, lift weights, bike, etc. I'd LOVE to imagine they used physio-lab methods (e.g., this report has some nice examples starting on slide 28) like direct or indirect calorimetrics during these activities -- which then allows the data from a device like the Band to be correlated with (1) the user's physical characteristics and (2) reasonably accurate (lab-based) burn estimates. Like I noted, this is all ignorant reverse engineering, or wishful thinking at best.

Another factor supporting my speculation is that the ambient temperature in these parts has varied wildly in the last month or two from 60 down into the negative digits. Interestingly, I've noticed on the exact same 1.5-mile walk I take with my dog, that the Band might say I burned 150 calories on a relatively warm day, versus over 300 today when it was in the single digits. It's clearly not the motion of my wrist alone that accounts for the difference -- not sure if it's HR or what, but the Band seems to think that when it's really cold outside, my body must be burning more calories, even though as far as I can tell, I'm taking pretty much the same walk each time. Again, that's more speculation.

tl;dr I'd like to imagine that the guys who program the Band are doing more than using conventional formulas to estimate calorie burn. It's a shame I don't have any real evidence to back up that hope!

​-Matt