It’s not the independent, peer-reviewed study that wearable watchers have been waiting for, but Healbe has published lab test data that indicates its GoBe activity-tracking wristband can automatically divine the calories in the food we eat with an error rate of +/-13.5 percent.
Should you believe the numbers? It all depends on how much confidence you place in relatively shallow reporting.
Healbe has the been the focus of intense scrutiny since April when it first announced its calorie-counting wearable. Medical and sensor experts say Healbe’s product claims are untenable, but now the company has published a brief Friday blog post that says its critics are wrong.
In the post, Healbe explains that it tested 30 volunteers over the course of 44 individual experiments. The test subjects included 11 men and 19 women across a wide ranges of ages. During the 16-day study, Healbe tested its subjects on “2-5 variations of a typical mixed menu, including breakfast, a snack, lunch, and dinner.”
It appears the actual caloric value of each meal was ascertained before consumption—Healbe says it compared food weight to established calories measurement tables. After each meal, Healbe checked the calorie consumption numbers reported by the GoBe wristbands, and cross-checked this information against data pulled from blood sample tests looking at blood glucose concentration.
When all was said and done, the company reports that its FLOW Technology algorithm was able to determine the calories that participants consumed within a 13.5 percent margin of error. Moreover, Healbe says, the actual types of macronutrients consumed has little effect on GoBe’s overall calorie report. It would seem you can eat any combination of carbs, proteins and fats, and this “affects the accuracy of the overall calorie measurement in only 3.5-11.5% of cases,” Healbe reports.
This would be quite a feat of algorithmic savvy considering that Healbe’s sensor appears to be looking entirely at blood glucose levels. Indeed, you could eat a 16-ounce steak—packed with calories to be sure—but those calories would be tied to proteins, not sugars. What’s more, medical experts have already told me that a simple impedance sensor (basically a sensor that measures fluid levels in body tissue) is incapable of accurate blood glucose concentration measurements.
“The physical reality is, this is just ridiculous,” Ries Robinson told me in March (you can read the full report here). Robinson has a medical degree from the University of New Mexico, a Masters in mechanical engineering from Stanford, and more than 20 years experience in developing systems for the optical measurement of body tissue. “It doesn’t work at a medical level. It doesn’t work at a practical level,” Robinson says of GoBe’s claims.
Nonetheless, Healbe is confidently, defiantly sticking to its original Indiegogo story that its $300 wearable can automatically determine the calories in the food you eat—all “through your skin, with no manual logging, no estimates, and no error-prone guesswork.” As Healbe managing director George Mikaberydze told TechHive in June, the system looks at blood glucose absorption and elevation rates, and then uses an algorithm to differentiate between fats, proteins and carbs. In the following video, Mikaberydze explains:
A fully peer-reviewed study—published in exacting detail—would eliminate skepticism surrounding this wearable that’s been skewered by myself, and absolutely savaged by PandoDaily’s James Robinson. But for now, all we have is a 470-word blog post that reports Healbe’s self-conducted findings, and doesn’t penetrate into any defensible level of detail. For example: What exact types of food were subjects eating? And why not publish raw data instead of a broad, top-line summary? And why not publish a quote from a medical expert that says, in effect, “I tested it, and it works!”?
I remain extremely skeptical of Healbe’s claims, but I’m also resigned to the likelihood that a relatively accepting consumer populace will be perfectly OK with a 13.5 percent margin of error should that particular claim prove to be legit. After all, today’s simple activity trackers report wildly different step counts, and can disagree by more than 20 percent when you’re wearing two different trackers on the very same wrist.
Yet the activity tracking wristband is still a product category vested with a large degree of legitimacy. Of course, if you think you’ve walked 12,000 steps when you’ve only walked 10,000, that’s not really a health risk. But if you’re obese and calorie tracking is paramount, wild disagreements between fact and fiction are a much more serious matter indeed.
Healbe says the GoBe will ship to Indiegogo backers in August, and will be available for sale in October. So stay tuned for my own independent report.
This story, "Healbe finally releases lab data for its notorious GoBe calorie-tracking wristband" was originally published by TechHive.