Explore self-monitoring with SIAL
When a calculator speaks for you
How does one keep track of one’s caloric intake and expenditure? There already exist more than enough gadgets that allow us to keep track of our physical data: nutritional scanners (TellSpec), activity trackers (Vessyl), smart clothing (SCiO), and connected flatware (like Baidu’s smart chopsticks) for low calorie kitchens.
The digital market is changing our relationship with food. From now on, we’re each equipped to be our own nutritional accountants. But for Marion Nestle, nutritionist and professor at New York University (NYU), the tracking of calories burned and consumed just isn’t viable:
“the lack of precision in self-reported food consumption [is] “legendary,” [with an] average underestimation of 30 percent.” (The Verge)
A journalist at Quartz who tested the FitBit Flex, the Jawbone UP24, the Basis, and the Misfit Shine found a difference of approximately 500 calories per day between these four activity trackers. The results are as follows:

Could a tendency towards blind faith in regards to gadgets be the culprit?
Not really, says The Verge, as there exists no rigorous scientific standard on the matter. As a result, the concept of scientific eating remains a complex subject.
Flash, analyze... and then maybe you can eat!
These variations in reported results happen to guide our daily dietary and fitness regimens. Barry Popkin, research professor at the Department of Nutrition at the UNC School of Medicine, is no stranger to this phenomenon.

According to him, these variations can cause users to take unnecessary health risks.
What’s more, these gadgets (like GoBe, which allows users to determine calories ingested and burned based on blood glucose levels) are at the centre of several debates on the viability of their tracking algorithms. Indeed, these algorithms are more often than not dubbed too generic to be taken seriously.
“The nutritional quantified self” : still too complex for our machines?
Just possibly. An example of this is Meal Snap, an application that uses image recognition technology in order to better inform users on the contents of their plates.
Meal Snap allows users to have some nutritional accountability by giving a calorie count based on only a photo of a meal, but the results show significant discrepancies.
TechCrunch, who tested the algorithm, reminded users that image recognition technology still isn’t accurate enough to identify all of a meal’s components (think: hidden fats). And even though the start-up uses Mechanical Turk to complete the results that Meal Snap alone can’t assess, there are still other issues to consider, like digestion.
These endless variations invite us to view technological gadgets not as an end in themselves but rather as a set of tools to assist proper medical accompaniment.
It’s worth noting, meanwhile, that research is making rapid strides to correct these defects. An example includes miniaturised molecular sensors which use spectroscopy to analyse food’s chemical chemical makeup (Airo Health). But here, too, doubts inevitably arise. To be continued in the next SIAL publication.