App users as good as trained experts at rating healthiness of foods
Crowdsourcing may help dieters stick to healthy foods and lose weight, as participants are as good as trained experts at correctly rating the healthiness of foods and giving feedback on them, indicates research published online in the Journal of the American Medical Informatics Association.
Being able to track energy intake and getting personalised feedback on diet have been linked to greater weight loss, but can be hard to sustain over time, say the researchers.
But the advent of smartphone apps, with the facility to upload photos in real time, which can be easily accessed on the go, could be about to change that, they say.
They wanted to find out how well crowdsourcing, which relies on the input of several smartphone users to provide feedback and information, might help people stick to a healthy diet.
They used 450 photos of food/drink uploaded onto the Eatery app by 333 unique users in Europe and the US.
This app enables users to rate their meals on a basic sliding ‘healthiness’ scale from ‘fit’ (healthy) to ‘fat’ (unhealthy), and to rate the photos of other app users in the same way, in a bid to help them improve the quality of their diet.
Three public health students, all of whom had completed coursework in dietary assessment, were asked to rate the same pictures, using a more complex scale, based on a set of nutritional standards – the 2010 US Dietary Guidelines – with points deducted for unhealthy components and added for healthy ones.
These guidelines aim to reduce dietary intake of foods high in salt, saturated and trans fats, sugar, and refined grains and alcohol, and increase consumption of fruits, vegetables, whole grains, low fat unsweetened dairy products and low cholesterol protein.
The expert raters’ scores were then compared with those of the app users. The results showed that both sets of ratings were similar, added to which the app users’ scores were in line with the national guidance.
They gave the ‘eat more of’ foods, such as fruits and vegetables, a higher healthiness score and the ‘eat less of’ foods, such as processed and fast foods and sugar sweetened drinks, a lower healthiness score.
The researchers say that their study represents the first step in assessing how reliable crowdsourcing might be for feedback on diet.
“Crowdsourcing has potential as a way to improve adherence to dietary self monitoring over a longer period of time,” write the researchers.
“The results of this study found that when basic feedback on diet quality by peer raters is crowdsourced, it is comparable to feedback from expert raters, and that peers can rate both healthy and unhealthy foods in the expected direction,” they conclude.
[The use of crowdsourcing for dietary self-monitoring: crowdsourced ratings of food pictures are comparable to ratings by trained observers Online First doi 10.1136/amiajnl-2014-002636]