Fitness trackers wildly off
A recent Stanford-linked thread going viral says consumer fitness trackers and gym treadmills can overestimate calorie burn — in some cases by as much as 93% — because they don’t account for individual fitness, efficiency, or device assumptions. (x.com) That matters because people use those numbers to justify extra food or lighter workouts, so the overestimates can easily erase a week’s worth of calorie balance if you trust them blindly. (x.com)
A calorie number on a watch is not a direct measurement like a bathroom scale reading. It is a guess built from signals like heart rate and motion, then run through an algorithm that tries to turn those signals into energy burned. (med.stanford.edu) The lab method for measuring energy burn is called indirect calorimetry, and it works by tracking oxygen in your breath while you exercise. Stanford’s 2017 validation study compared seven wrist devices against that lab standard in 60 adults during sitting, walking, running, and cycling. (stanfordhealthcare.org) The watches were pretty good at heart rate. Six of the seven devices had heart-rate error below 5 percent, which is why pulse tracking usually feels believable even when calorie tracking does not. (med.stanford.edu) Calories were the weak spot. In the same Stanford study, no device got energy-expenditure error below 20 percent, the best device was off by 27 percent on average, and the worst was off by 93 percent. (med.stanford.edu) The reason is simple: two people can do the same workout and burn different amounts of energy. The Stanford paper found error changed with sex, body mass index, skin tone, and activity type, with higher error during walking than cycling. (stanfordhealthcare.org) Wrist motion is also a noisy clue for calorie burn. Stanford engineers said in 2021 that smartwatches and phones are often off by about 40 to 80 percent because wrist swing and heart rate do not reliably map to true energy use. (news.stanford.edu) That is why a hard ride and an easy ride can confuse a device in different ways. Heart rate can rise from caffeine or stress, and wrist movement can look busy even when the legs and lungs are not doing much extra work. (news.stanford.edu) The treadmill number at the gym has the same basic problem. Most machines estimate from speed, incline, time, and a default body profile, so they miss differences in running economy, fitness, stride, handrail use, and whether the user entered accurate weight data. (marathonhandbook.com) Large reviews have landed in the same place as the Stanford experiment. A British Journal of Sports Medicine systematic review found wearable estimates of energy expenditure were highly variable across 60 studies, with accuracy depending on device and activity type. (bjsm.bmj.com) A 2020 systematic review covering 158 publications and nine brands also found commercial wearables were much better at steps and heart rate than at energy expenditure. The review grouped calorie estimates with the weakest validity across the major metrics consumers actually use. (mhealth.jmir.org) Stanford’s answer in 2021 was to move the sensors off the wrist and onto the leg, because leg motion tracks the work of walking and running more directly. That prototype averaged about 13 percent error, versus roughly 40 to 80 percent for typical smartwatches and phones. (news.stanford.edu) The practical takeaway is not that every watch is useless. It is that a calorie number from a watch or treadmill works better as a rough trend line over weeks than as permission for a 500-calorie dessert after one workout. (med.stanford.edu)
