By Brian X. Chen, The New York Times Company
Many of us are in the same boat these days. With the coronavirus killing more people by the day, we are increasingly stress-eating and drinking more alcohol. At the same time, with gyms shut down, we are sitting around more and glued to screens.
So you may be wondering what I’m wondering: How is the pandemic affecting my body? Because we can’t easily leave the house to see doctors for nonemergencies, we are largely left to figure this out on our own.
Enter the Halo, a new fitness-tracking bracelet from Amazon with a novel twist: It claims that by using a smartphone app to scan images of your body, it can tell you how much body fat you have much more precisely than past technologies. The bracelet also has a microphone to listen to your tone of voice and tell you how your mood sounds to other people. (The masochist inside me said, “Sign me up!”)
The Halo is Amazon’s foray into so-called wearable computers that keep an eye on our health, following in the footsteps of Apple and Fitbit. Amazon is selling the Halo for $65 on an invitation-only basis, meaning you have to get on a waiting list to buy it. I volunteered to be a guinea pig and received mine in October.
When the Halo arrived, I installed the app, removed my T-shirt and propped up my phone camera. Here’s what happened next: The Halo said I was fatter than I thought — with 25% body fat, which the app said was “too high.”
I was skeptical. I’m a relatively slim person who has put on 2 pounds since last year. I usually cook healthy meals and do light exercises outdoors. My clothes still fit.
I felt body-shamed and confused by the Halo. So I sent my Halo data and body scans to Dr. Lawrence Cheskin, a professor of nutrition and food studies at George Mason University and founder of the Johns Hopkins Weight Management Center.
After reviewing my results, Cheskin jotted down my height and weight to calculate my body mass index, which is a metric used to estimate obesity. A man my age (36) with my body mass index, he said, is highly unlikely to have 25% body fat.
“Unless you were a couch potato and ate a very poor diet, I have my doubts about the Halo’s diagnosis,” he said.
Cheskin encouraged me to gather more data by measuring my body fat with other devices, and to do the same with at least one other person. So I did and found that the Halo’s body fat readings consistently skewed higher than other tools for myself and my test subject.
I concluded that the Halo’s body analysis was questionable. More important, it felt like a negative experience that failed to motivate me to get fit. I’ve had much more uplifting experiences with other products like the Apple Watch and Fitbit bands, as laid out below.
Measuring body fat
Body fat measurement can be complicated because the traditional methods available to consumers are not always accurate.
Smart bathroom scales that measure body fat use bioelectrical impedance analysis, which sends a small current through your bare feet. Skin calipers, a more dated method, are essentially rulers that pinch down on skin folds to measure thickness.
These techniques are not perfectly reliable. If people step on smart scales at different times of day or with different levels of hydration, their results may vary. Calipers can measure skin folds incorrectly if you pinch in the wrong areas.
Amazon said the Halo’s technology was much more precise. To scan your body, you use the smartphone’s front-facing camera to take photos of your body from the front, sides and rear. Then Amazon stitches the images together into a 3D model to analyze your body composition and calculate the percentage of fat.
I decided to record consistent body fat measurements for myself and a friend using the Halo, a Fitbit bathroom scale and a highly rated skin caliper. In November and December, I took early-morning measurements with the Halo and bathroom scale; my wife pinched my skin folds in four areas with the caliper. I measured my test subject’s body fat once with each device.
Our results were remarkably similar for two men with very different body compositions:
— The Amazon product estimated that my friend, a 6-foot-3 man weighing 198 pounds, had 24% body fat, the Fitbit scale read 19%, and the skin-fold measurements added up to 20%.
— For myself — 5-foot-6 and about 140 pounds — the Halo said in November that I had 25% body fat, the Fitbit scale said 19%, and the skin-fold measurements added up to 20%. In December, the Halo said I had 26% body fat (alas, I had more Thanksgiving leftovers than usual), the Fitbit scale said 20%, and the skin-fold measurements added up to 21%.
Cheskin speculated that the Halo might have an overestimating bias in its algorithm because underestimating body fat for an obese person would be more problematic.
Dr. Maulik Majmudar, Amazon’s medical officer, who worked on the Halo, said people should expect the device’s results to be different because the method was more accurate than body fat scales and calipers.
Amazon developed its body-measuring algorithm from a sample set of tens of thousands of images of people’s bodies from across a wide range of demographics, he said. Amazon then did internal tests measuring people’s body fat using the Halo scanner, smart bathroom scales and DEXA, a technique that uses X-rays to scan for bone density, which studies have found to be a reliable measure for body fat. It found that the Halo method was twice as accurate as bathroom scales.
Still, Cheskin was unconvinced by Amazon’s accuracy claims. He said a valid study would involve a clinical trial measuring body fat of many human subjects with each method — the Halo, DEXA, bioelectrical impedance scales and calipers — and comparing the results side by side.
Accurate or not, the most disappointing part of Amazon’s body fat analysis was that it lacked important context. Even though the app asked for my ethnicity, age and sex, it said my 25% body fat level was too high and well outside the “Healthy” zone (roughly 12% to 18%). It also said healthy results were associated with longer life and lower risks of heart disease.
Cheskin offered a more nuanced analysis. Body fat levels may have different health implications depending on your age, ethnicity, sex, cholesterol levels and family history. Waist circumference matters, too, because severe abdominal fat can be associated with health problems.
For an Asian man my age with a 34-inch waistline, whose family has not had a history of diabetes or heart problems, and whose blood tests recently showed normal cholesterol levels, even a 25% body fat reading would probably not be alarming, he said.
That context, combined with my body mass index along with the measurements taken with a body fat scale and caliper, led Cheskin to doubt Halo’s analysis.
He worried about the technology’s potential consequences.
“Does it potentially create eating disorders?” he said. “You’re taking a bunch of people with normal weight and BMI and telling them they’re too fatty. What are they going to do with that? Some of them are going to be more compulsive and start doing things that are going to be inappropriate.”
In my experience, there are better fitness-tracking products that offer more positive motivation.
The Apple Watch, for one, lets you set goals for how much you want to move or exercise each day, and those goals are symbolized by colorful rings that are shown on the watch face. Once a ring is completed, you have met your goal. Fitbit devices send notifications to your phone, egging you on when you are nearing your step goal. Neither device comes anywhere close to giving you body dysmorphia.
Another of Halo’s unique features is Tone, which uses the bracelet’s microphone to periodically listen in on your conversations to tell you what your mood sounds like. I turned the feature off after two days because it felt like a creepy invasion of privacy. But I left it on long enough to complain to my wife about what a bad idea it was.
After analyzing the conversation, the Halo app said I sounded irritated and disgusted. That, at least, was accurate.