At the National Institutes of Health, a group of adult volunteers participated in a tightly controlled dietary study. Some served meals composed of industrially processed items—pre-packaged waffles, processed meats, and artificially sweetened beverages—while others received freshly prepared, minimally processed foods. But rather than relying on participants’ descriptions of their meals, researchers focused on analyzing biological samples collected afterward, aiming to uncover what the body itself could reveal about dietary intake.
In their new study, the NIH researchers describe their novel method to detect how much ultra-processed food — or UPF — a person consumes, simply by analyzing their blood and urine. This marks the first time that scientists have identified biological fingerprints that reliably reflect an individual’s intake of industrially manufactured foods.
Ultra-Processed, Ubiquitous — and Unmeasured
From sugary cereals and sodas to factory-made bread and frozen pizzas, ultra-processed foods now dominate diets across the globe. In the United States, they account for nearly 60% of the average person’s calorie intake. These foods are engineered for shelf-life, flavor, and profit — often at the expense of nutrition.
But scientists have struggled to measure just how much of these foods people actually consume. Most studies rely on food diaries or dietary recall — tools as unreliable as memory itself.
“There’s a need for both a more objective measure and potentially also a more accurate measure,” Erikka Loftfield, an epidemiologist at the U.S. National Cancer Institute and lead author of the study, told The Associated Press.
To address that gap, Loftfield and her colleagues turned to metabolites — molecules produced as our bodies digest and process food. These tiny chemical traces linger in our blood and urine, offering a molecular snapshot of what we’ve been eating.
In a study involving 718 adults aged 50 to 74, researchers collected blood and urine samples from each participant twice over a year. Participants also provided detailed food logs. The scientists then used machine learning to analyze over 1,000 metabolites, searching for patterns that aligned with ultra-processed food consumption.
What they found was striking. People with higher UPF consumption — sometimes as much as 82% of their daily calories — had distinct metabolic signatures. Their samples contained more markers linked to type 2 diabetes and even traces of chemicals associated with food packaging. Conversely, they had fewer metabolites from fresh fruits and vegetables.
Your Metabolism Doesn’t Lie
To test their method, the team drew on data from a tightly controlled study at the NIH Clinical Center, published in Cell Metabolism. Twenty volunteers lived on-site and ate only what researchers gave them — two weeks of a diet rich in ultra-processed foods, followed by two weeks of whole foods, or vice versa.
The researchers could tell, with precision, when each participant switched diets — not by what they reported eating, but by what their blood and urine revealed.
“We found this signature that was sort of predictive of this dietary pattern that’s high in ultraprocessed food and not just a specific food item here and there,” Loftfield said.
From these results, the team created what they call a “poly-metabolite score.” With just 28 blood markers or 33 urine markers, scientists could estimate how much energy someone derives from UPFs.
The Road Ahead
The study wasn’t meant to prove that ultra-processed foods cause health problems. It also doesn’t pinpoint which ingredients — emulsifiers, preservatives, artificial sweeteners — might be the true culprits.
Still, the ability to objectively measure UPF intake could help untangle these threads. With this tool, researchers could revisit old studies where biological samples were saved but diet was poorly documented. They could track long-term health outcomes — including cancer — with unprecedented accuracy.
If metabolite scoring is validated in larger, more diverse populations, it could usher in a new era of nutrition science — one that listens less to what people say they eat, and more to what their bodies reveal.
“We’re sort of trapped in this industrial food-production system where we all eat ultra-processed food,” Robinson told Nature. “It’s quite hard for most people to go back to fresh food, because our food systems are now set up like this.”
And yet, thanks to a few drops of blood and urine, we may finally have a way to see what that diet is doing to us — molecule by molecule.
The findings appeared in the journal PLOS Medicine.
