A 12-Week Study of the Accuracy of the Apple Watch’s Calorie Tracking

Tracking calories while swimming with the Apple Watch

In the fall of 2019, I conducted a 12-week experiment to determine how accurately the Apple Watch calculates ?energy expended? (EE), both while resting and while exercising.

There have, of course, been a number of studies attempting to quantify the accuracy of smartwatch calorie tracking, but as far as I can tell they were mostly short-term (a few hours), and didn?t measure the correlation between EE estimates and weight. That is, they don?t answer the question ?Can I use the calorie estimates from the Apple Watch to lose weight?? I think the answer, for many people, is ?yes?.

Here?s what I did:

1. For 84 days, I tracked all the calorie-containing food I ate, and some of my diet soda consumption, using a food tracking app and an Alexa skill that I wrote for this experiment. I wrote my own app because I needed quality-controlled calorie counts for the foods I was eating, and very fast entry of foods, which is a necessity when trying to track every calorie that one consumes (I tracked amounts as small as a few jellybeans).
2. I wore my Apple Watch Series 5 as much as I could, approximately 22 hours a day. I only took the watch off to charge it.
3. I weighed myself with a digital scale that also estimates body fat percentage. I know the scale is consistent measuring weight (I tested it using objects of known weight), but I have no idea how accurate the body fat measurements are. I also have no idea if iOS or watchOS uses body fat measurements for its EE calculations.
4. All of the data for this experiment can be found in this Google spreadsheet, including every food I ate and the portion size (there are approximately 1,300 entries, an average of 15.5 per day).

The results are quite clear in the graph below. The correlation between the predicted calorie deficit (according to the Apple Watch) and my actual weight is very good. After 79 days, the calorie deficit using the EE numbers from the Apple Watch was 87,157. Assuming 3,500 kcal/lb., my weight should have been 193.7 lbs. My actual weight? 193.3 lbs.

What if a person has a slower (or faster) metabolism, and needs a deficit of, say, 4,000 kcal to lose a pound of fat? Weight loss would be slower, but the correlation would be the same.

Click image for interactive version

More Details

I?m a 49 year-old male, 5’9″, with no pre-existing conditions other than elevated blood pressure of 140/85. At the beginning of this experiment, I weighed 218 pounds (obese), and at the end I weighed 193 pounds (no longer obese, just significantly overweight). Weight loss of two pounds a week is fast, but not excessive.

As for my diet, it isn?t very good, as you can see by perusing the food diary. It?s a lot of processed food, fast food, and too much diet soda. There are more White Castle sliders on the list than salads, unfortunately. The only advantages to eating like this were:

• It more closely approximates the diet of the average overweight American than if I?d gone on one of the currently popular fad diets.
• It made it easier to track my calories exactly, as processed food has clearly marked nutritional information.

I exercised moderately during the period of the experiment, mostly swimming laps and walking. On about half the days, I ?closed my rings?.

I used the ?Activity? app on iOS to determine my energy expenditure (circled in blue)

Conclusion

I?m frankly surprised that the EE numbers from the Apple Watch are this good. All the usual caveats apply: this was only one person, I may not be typical, my activity pattern might be well suited to being tracked by the Apple Watch, etc. But I was very rigorous in how I collected this data, and I?m now much more confident in using the Apple Watch as part of my weight loss regimen.

What?s next? Now that the holidays are over, and I go back to tracking my diet, I?m going to repeat the experiment while trying to reach my goal weight of approximately 170 lbs. If I?m successful in reaching that goal, the second phase of the experiment might be even more interesting because my body composition (fat vs. muscle) will change more than it did in the first 12 week experiment.

I can be contacted at [email protected]. If you are a researcher, I can provide full exports of my diet, including exact times when food was consumed and every nutrient in those foods (warning: this is over 10,000 data points), as well as a data dump from my digital scale and my Apple Watch.

I have no association with Apple, Inc. and I paid full retail price for my Apple Watch.