Northeast experts take the mystery out of what we eat.
In an article published Tuesday in Nature Food, Julia Menihettisenior researcher at Northeastern's Network Science Institute, demonstrates that concentrations of different nutrients in foods follow a consistent pattern and that the amount of any given nutrient in a food follows a similar mathematical formula.
Inspired by these findings, her team found that 73% of the US food supply is over-processed, which is linked to a higher risk of developing a variety of health problems. Their findings, which showed the level of processing for more than 50,000 foods sold in the United States, were published in an online database for public use.
Menichetti hopes this database will help fill gaps in the public's knowledge about what they're eating—specifically, how processed the food actually is.
“It's really food and environment, not just genetics, that are the main determinants of our health,” he says Giulia Menichetti, Senior Research Scientist at Northeastern's Network Science Institute. Photo by Matthew Modoono/Northeastern University
The study is part of a larger project called Foodwhich he co-founded Albert-László Barabási, Robert Gray Dodge Professor of Network Science at Northeastern and co-author of the study. Similar to what the Human Genome Project did for human genetics, Foodome seeks to map all the chemical components of the human diet, with the goal of better understanding how what we eat affects our health.
“It's really food and environment, not just genetics, that are the main determinants of our health,” says Menichetti.
That's good, he says. Unlike our genetic makeup, diet is something that humans have control over. But to do that requires knowing what's actually in our food. That's why Menichetti's team is mapping food's “dark matter”—or the unknown chemical components beyond what's listed on nutrition facts. This could help us understand what is in what we eat and how much it has been processed before it reaches our plates.
What does it mean for a food to be processed — and why does it matter? Barabási says that processing applies to literally anything you do to a food, like chopping vegetables.
“That in itself is not a problem,” he says. “The problem is over-processing.”
For a food to be ultra-processed, he says, it must have been chemically altered. An example is some orange juices that are labeled “natural” but are actually separated into three different chemicals before being stored separately and blended later.
There may be no indication on a package that this product is ultra-processed, he says. The USDA only monitors and reports so many nutrients, and the FDA only requires companies to report about 12 nutrients. This is a problem because, the group claims, there is a link between highly processed foods and a “higher risk of metabolic syndrome, diabetes, angina, elevated blood pressure and biological age.”
Albert-László Barabási, Robert Gray Dodge Professor of Network Science at Northeastern. Photo by Ruby Wallau/Northeastern University
Menichetti's findings take a step toward a better understanding of all food chemicals. In her work, she observes that natural nutrients exhibit common patterns, which are well captured by an equation.
The finding is unprecedented, says Barabasi. “The very existence of this formula was the most amazing thing,” he says. “No one has even realized this is possible.”
Unfortunately, there is no biomarker or chemical marker for highly processed foods. But in two follow-up papers, both of which are under review, Menichetti shows that by revealing what nutrient concentrations should look like in natural, unprocessed foods, the equation can help us determine which foods in the U.S. supply were chemically altered and therefore deviated from the spectrum of nutrients seen in natural ingredients.
“It suggests a way to detect things that are outliers, things that don't behave in the ranges seen in natural ingredients,” he says. “What we're seeing is that highly processed foods … basically behave in a way that shows extreme concentrations of many different nutrients.” For example, when an onion is fried and pounded, more than half of its nutrients change in concentration. these changes correlate with the processing level.
Knowing this, the team set out to figure out how much of all US food is over-processed. Thanks to machine learning, Menichetti and her co-authors—Babak Ravandi, a postdoctoral researcher at Northeastern and Dariush Mozaffarian of Tufts University—were able to do just that.
Their final task was to share this information with the public so that people can make more informed decisions about their diet. The third paper, which Menichetti co-authored with Ravandi and research assistant Peter Mehler, introduces GroceryDB, a database that includes information on more than 50,000 food items sold at major grocery retailers.
The online edition of the database allows consumers to browse the food supply for the level of processing. Each food is given a score from 0 to 100 and users can compare different products. For example, Triscuits with a touch of sea salt have a score of 89, original Cheez-Its a score of 57, and whole wheat Ritz crackers a score of 29. Meanwhile, reduced-fat Wheat Thins received a 3.
For media inquiriescontact Marirose Sartoretto at m.sartoretto@northeastern.edu or 617-373-5718.
