Amazon’s Halo body fat percentage calculator outperforms lab devices

Amazon’s Halo

Amazon’s body fat percentage scanner performs better than other, more cumbersome methods of calculating body fat, a new study showed. A body scan is one feature on the Amazon Halo Band subscription service, which the company announced last August.

The Halo Body feature works through smartphone cameras. Users take four photos of their bodies, which are then combined into a 3D image. Then, Amazon’s tool uses machine learning to calculate the user’s body fat percentage. When Amazon launched Halo, the company said it did an internal study to validate the feature. Now, they’ve released a study — which was funded by Amazon, but conducted with leading experts in the field — showing that the tool works well.

The results show that the tool could be a benchmark for people looking to get more information about their bodies, says Diana Thomas, a mathematician who studies body weight regulation at West Point and was not involved in the study. She wouldn’t use it for scientific research, but it could give users another data point. “It gives you more information,” she says. “If you have a high percent body fat flagged by this, it could be a red flag for you.”

Body fat percentage can be a more accurate way to assess someone’s health than their weight. Two people at the same size could have different amounts of fat relative to their overall weight and would have different health risks as a result. Having some body fat is important, but having a high percentage is associated with health problems like heart conditions and diabetes. The percentage of body fat considered healthy varies for people of different ages and sexes. Women, for example, have more body fat than men.

The study used data collected at two sites: Massachusetts General Hospital and Pennington Biomedical Research Center at Louisiana State University. It had 134 participants of varying body types and included 82 women and 52 men. Of the group, 81 were Caucasian, 32 were Black, nine were Asian, four were Hispanic, one was American Indian, and seven were multiracial. It’s a small number of people, but that’s typical with this type of research, which has to run people through expensive and unwieldy tests, Thomas says.

Each person in the study had their body fat percentage calculated through an X-ray technique that shows fat distribution through the body called dual-energy X-ray absorptiometry (DXA), which is the gold standard. The results from those scans were compared to the Amazon system, five scales that calculate body fat (two professional and three consumer), and something called air displacement plethysmography — which uses a pod-shaped device that measures how someone’s body displaces the air.