The calories burned during some activities have gotten increasingly inaccurate. I track my runs on another app which, in the past, normally lines up with my Garmin. The Garmin calories have been steadily decreasing and today it showed half the number of calories burned as my other app. It shoes I burned 400 calories on a 9.5 mile run. The calories for cardio workouts have also been inconsistent. What can I do to fix it?
OK, here you go. It took me an hour of my free time, but what wouldn't I do for a fellow athlete. Hope you appreciate it, at least.
I entered the data of all three activities, and the resting HR for the respective days, into an Excel table. I do not know the exact algorithm from Firstbeat that Garmin uses for their Calories calculations, so I used a commonly available formula for calculating Calories from HR (source: Quora). I just did a minor change, since I do know that the Firstbeat algorithm uses the resting HR, hence I implemented it into the formula too. Then I took just a wild guess, and did the calculation for a female athlete 30 years old and weighting 50kg.
The resulting Calories using the common formula from Quora are in the column calc Cal, and the Calories reported by Garmin Connect are in the column GC Cal. The ratio calc/GC is then in the last column.
As you can see yourself, the values match almost perfectly for all three cases (margin of error of only -1% to +2% !). Frankly told, I did not even expect that perfectly matching results, because I believe the calculation Garmin does, is much more complicated, using aso the time in individual HR zones. They also certainly do not use the average HR for the entire activity time, but rather aggregate the energy expenditure over the entire time. That explains the slight difference of the results from the simplified algorithm I used.
The conclusion is simple and clear: Garmin is right, your other app has it wrong. And that for the simple reason that they most likely do not calculate with the measured effort (the HR), and only estimate the Calories from the distance and time, instead.
PS: The formula I used is shown below, including the RHR inclusion (41 + [@[Δ HR]])). It certainly needs some tuning to match better the true age and weight of the athlete, but I am not really willing to spend more time on it. I believe that as the evidence it is more than sufficient.
((-20,4022 + (0,4472 * (41 bpm + ΔHR)) - (0,1263 * 50 kg) + (0,074 * 30 years))/4,184) * 1440 * activityTimeInMin
