Can our keywords reveal our eating habits?
Analysing the keywords we use when we search Google is not a new thing. Google Flu Trends is the most popular example, using keyword analysis to determine the spread of flu. Could it also provide insight into our eating habits?
That’s the question asked by some new research from Stanford and Microsoft. They wanted to determine whether the searching we all do for recipes, demonstration videos and even our online grocery shopping can reveal the shifting eating habits of entire communities.
The research unveiled some interesting insights into how our diets are changing. For instance important trends emerged regarding the way our diets change according to our geographic location or on special days, such as Christmas. The researchers believe this will have some important implications for healthcare professionals and policy makers.
Their research consisted of analysis of the recipes people searched for over a 16 month period. The sample consisted of people from around the world who had downloaded a browser plug-in for Bing that consented to the use of their search data. They combined what people were searching for with a detailed breakdown of the nutritional value associated with recipes.
The research unveiled some clear seasonal variations, with summer being a period of relative fasting compared to the feasts we enjoy over winter. “Overall caloric intake is lowest in summer (July and August) and peaks in fall and winter,” say the researchers, with the difference being about 30 kcal per serving.
This is clearly a seasonal thing, because the trend is reversed for people in the southern hemisphere. So for them, a summer Christmas means eating less than it does for northern hemisphere folks that enjoy festivities during winter.
They also say that the average caloric intake per serving is signi?cantly lower in the southern than the northern hemisphere at 285 versus 300 kcal, although they offer no explanation for this.
Interestingly, there are also some clear trends regarding what we eat throughout the week. Calorific intake tends to peak on Tuesdays, before reaching a low point on Fridays. A rather predictable spike in calorie consumption is seen on special events, such as Christmas day or Thanksgiving.
All of which is interesting, but how useful are these findings? Of course it’s important to remember that the data is purely for what we search for rather than what we actually eat. Whilst it’s likely that we end up making what we search for, it is not guaranteed, so that remains an important caveat to the research.
One could also argue that the people searching for recipes online are a relatively small sample of the overall population. Are such people sufficiently representative of the norm to provide us with insight into the eating habits of an entire nation?
Whilst the researchers believe that monitoring this data could provide healthcare professionals with a real-time insight into the eating habits in their communities, thus allowing them to target messages at people at the optimum times, I think a bit more research would need to be done into the accuracy of the data and the trends that have emerged from it. It does certainly represent an interesting new angle in the ongoing battle to keep our waistlines in shape.