Early trial showed use of computer algorithm to produce diet tailored to a person’s unique biological make-up had benefits for pre-diabetic subjects

Bifidobacteria, a type of gut bacterium. Researchers believe that the different types of gut microbes found in individuals has a significant impact on how their bodies respond to food. Photograph: Phototake Inc./Alamy

Scientists have created bespoke diets using a computer algorithm that learns how individual bodies respond to different foods.

Researchers believe the tailored diets could help stem the rising tide of diabetes, heart disease and obesity, by personalising people’s daily meals and so helping them to adopt healthy eating habits.

The first results from the Personalised Nutrition Project, run by leading researchers in Israel, are due to be unveiled on Friday at the Human Microbiome conference in Heidelberg, Germany.

The project challenges the idea that general recommendations about healthy foods are suitable for everyone, and instead aims to produce optimised diets based on people’s unique biological make-up.

“We are all different,” said Eran Segal, a computational biologist who runs the project with Eran Elinav at the Weizmann Institute in Rehovot. “We see tremendous variability in people’s responses to foods, so if you want to prescribe diets, they have to be personally tailored.”

An early trial, run by Segal and Elinav, found that tailored diets – designed with the computer algorithm – benefited 20 people with pre-diabetes, by preventing high spikes in blood glucose levels after meals. Some of the participants found their blood glucose had returned to healthy levels during the course of the study.

“In the end, I think this will be relevant for everyone, whether they want to lose weight, or to maintain their weight,” Segal told the Guardian ahead of the conference.

The scientists have enrolled nearly 1000 people for the project over the past two years. In the first phase of the project each person wore a blood glucose monitor and kept a diary of their eating habits and lifestyle for a week. The monitor recorded their blood glucose levels every five minutes, revealing how their bodies responded to different foods.

When a person eats a meal, the food is broken down into sugars in the gut. Most of that sugar is glucose and passes straight into the bloodstream. Because high levels of blood glucose can be harmful, the body releases insulin. This makes cells absorb the circulating sugar, which use it for energy or store it as fat.

When the body can no longer control blood sugar levels it can suffer a host of metabolic diseases, including diabetes, which are on the rise despite healthy eating guidelines, such as recommendations to eat five portions of fresh fruit and vegetables a day.

Segal believes that personalised diets could be more effective, and help people to control their blood sugar and prevent them developing diseases. “Blood glucose is key to weight management and diabetes, and is linked to many, many other diseases, including cancer,” he said.

The scientists gathered data on how people responded to around 50,000 meals, and found that they varied enormously. “In some people, when they have bread, they show no change in glucose levels, but others spike dramatically,” Segal said.

To investigate why people were so different, the researchers looked at the make-up of the participants’ microbiomes: the rich collection of bugs that lived in their guts. They found that the types of microbes they harboured had a significant impact on how their bodies responded to meals.

In the last part of the study, the researchers trained a computer algorithm on their data and found that it could accurately predict how different people would respond to particular meals.

To test the algorithm, Segal and Elinav used it to draw up bespoke diets for 20 pre-diabetic people. In the small trial each patient was given two very different diets. In the first week, their diet was tailored to minimise spikes in their blood glucose levels. In the second week, their diet was designed to have the same calorie content, but not to control their blood sugar.

“In all these cases, there was a big difference between the good diet and the bad diet, even though they contained the same calories,” said Segal. “By personalising these diets, on the good week, in some people, blood glucose fell to healthy levels, whereas in the bad diet week, they had glucose spikes that would be considered as glucose intolerant.”

The study highlights how traditional thinking around diets is flawed in the assumption that people put on weight purely because their meals contain more calories than they burn off. “Calories are definitely an important player, but we’ve been led to think that it’s the only player, and that is absolutely not true,” said Segal.

In many cases, he said, the tailored diets included foods that some might find surprising. For example, many people had low glucose responses to ice cream, and bread and butter tended to trigger less of a glucose response than unbuttered bread. “There are many more such surprises, including foods considered to be good which on average are not,” he added. “Our entire approach is data driven, not based on hypotheses or preconceptions, which in our view makes it powerful and science based.”

The researchers have yet to publish their findings in a peer-reviewed journal, so the results are tentative for now. But in the next few months, the scientists will launch a much larger trial to see how effective tailored diets might be.

Tim Spector, professor of genetic epidemiology at King’s College London, and author of the 2015 book, The Diet Myth, said that studies have found even identical twins to respond differently to similar diets, and that a difference in gut microbes was partly responsible. “This whole idea of personalised diets is the direction that we are moving in,” he said.

A major question is whether people are likely to stick with computer-tailored diets more than they do with others. Spector believes that might, if the diet is tailor-made from a wealth of scientific data on the person. “If you give people extra reasons for why they should do things, they are more likely to comply,” he said.

Yuval Dor, a professor of biology at the Hebrew University of Jerusalem, who has discussed the results with Segal, but is not a collaborator, said the work had tremendous potential. “This may open up new ways to design nutrition to control the outcome much better,” he said. “It could be of huge value for pre-diabetics as well as for people with both type 1 and type 2 diabetes” who have to control their blood sugar levels. “Eran may come up with an entirely new, simple and feasible way of achieving this,” he said.