Over the years, several different food-rating systems have been devised to help people with diabetes decide what they should, and shouldn’t, eat. Amy Campbell recently covered some of these systems in her blog here at DiabetesSelfManagement.com — including carb counting, which involves keeping detailed track of carbohydrates in foods; diabetes exchanges, which involve eating a certain number of “choices” in different food categories at each meal; and the glycemic index, which aims to predict how certain foods will affect a person’s blood glucose level.
But what if you could log your meals, activities, and blood glucose readings for a week using a smartphone app, and a computer algorithm could tell you what to eat? That’s the vision that was recently put to the test in an Israeli study.
Published last week in the journal Cell, the study looked at one other key piece of data: the mix of bacteria found in a person’s stool sample. The researchers speculated that differences in natural gut bacteria might help explain why some people have vastly different reactions to the same food, in terms of how it affects their blood glucose level.
As noted in a HealthDay article on the study, individual differences in reactions make systems like the glycemic index useless in many cases. After eating bread, some study participants had significant blood glucose spikes, while others had almost no reaction. When butter was added to the bread, some participants saw even greater blood glucose spikes, while in others blood glucose spiked less sharply than without the butter. In one woman, tomatoes seemed to cause unusually large blood glucose spikes.
After collecting all of the smartphone-logged information from participants — along with profiles of their gut bacteria, and basic information like sex, age, and weight — the researchers designed a computer algorithm to predict what physical characteristics (including gut bacteria) and behaviors might account for different blood glucose reactions to different foods. Then, they created individual diets for 26 participants based on this algorithm, and had them follow a “good” diet — according to the algorithm — for a week, as well as a “bad” diet for a week.
Not surprisingly, participants had lower postmeal blood glucose levels when they followed the “good” diet. More surprisingly, for some participants a “good” diet included foods, like pizza and potatoes, that tend to spike blood glucose levels in most people.
What do you think about this approach to personalized nutrition advice — would you be interested in such a program, if it were available to you? Would you feel strange giving a lab a stool sample to help inform your dietary choices? Do you think a program like this will ever be widely available? Have you ever noticed that you react differently to certain foods, in terms of your blood glucose level, than other people? Leave a comment below!