Obesity and type-2 diabetes are at frightening levels worldwide. Losing weight is the best way to help combat this rising epidemic. There are many diets claiming to help you lose weight quickly and many companies that have made millions by convincing people that their diet was the one that would work. However, the fact is most diets don’t help the majority of people lose weight. What works for one person is almost certainly going to fail for someone else, why is that? Researchers out of Israel have published a study in Cell looking at the possibility of personalizing a diet for a person by measuring how their environment affects the rise in glucose (sugar) in their blood following a meal. There is much research going on that shows excessive sugar intake and therefore blood sugar levels could be one of the causes of the obesity epidemic and is likely the cause of the type-2 diabetes problem we face.
From 800 participants the researchers collected a number of data and samples including: age, sex, height, weight, foods eaten, medication taken, level of activity, disease status, bacterial colonies in the gut, and blood. They also continuously monitored the blood glucose levels by using a sensor that goes under the skin. By collecting all the variables and plugging them into a computational model, the researchers were able to build a system that could predict which foods a person could eat to keep their daily blood glucose levels at a minimum. They also were able to determine which factors in a person’s life could predict how they would respond to a certain food or diet.
For example, two people ate a banana and a cookie. Person A had a spike in blood sugar 30 minutes after eating the cookie, but not after eating the banana. Person B ate the banana and had a blood sugar spike 30 minutes later but didn’t respond the same way to the cookie. The researchers saw that variables like HDL cholesterol levels (good cholesterol), hip circumference, and blood pressure was related to whether the cookie would cause a spike in your blood sugar.
Additionally, the researchers were able to take their computational model, apply it to a group of 26 individuals, and randomize them into diets that should either be good for them (based on the algorithm inputs) or bad for them. When they did this, they saw that the algorithm was able to predict which foods a person should eat to minimize their blood glucose levels and which foods would result in spikes in blood glucose. Sometimes, foods on one person’s good list were on another person’s bad list. For example, one person had their blood sugar levels spike when they ate tomatoes while for other people tomatoes were on their good list.
So what can be done with this type of information. Surely we wont be collecting hundreds of personal measurements on each person (some of them invasive) just to nail down the right diet for a person. The neat thing about this algorithm is that you should be able to whittle it down to only a handful of variables that can reliably predict someone’s response to certain types of food. Using this information, someone trying to lose weight to better their health could leave a doctor’s office knowing which foods they should avoid. This type of algorithm is still some time away from market use but it shows the importance of individual factors that determine how you respond to not only food, but potentially also treatments.