Your Body Keeps the Score: How a New Biomarker Could Change the Way We Understand Ultra-Processed Foods

When Erin retired, she did what many health-conscious retirees do—booked a full physical check-up. Her doctor handed her the lab results with a calm voice and a raised eyebrow. “Your cholesterol is creeping up, and your blood sugar is flirting with the line. Maybe we should take a closer look at your diet.”

Erin nodded, a little puzzled. She didn’t think there was anything wrong with the way she ate. No fast food, no sugary sodas. Breakfast was a protein bar and skim milk. Lunch? Grilled chicken breast. Dinner? One of those organic veggie microwave bowls from the health aisle. She’d been careful. Or at least, she thought she had.

But doctors have learned something over the years: patients like Erin are everywhere. People who follow labels and trends, who steer clear of “junk food,” yet still end up on the slippery slope toward chronic disease. The problem isn’t necessarily what they ate—it’s what they believed they ate.

This gap between perception and reality is where ultra-processed foods quietly thrive.

Most of us think we’re safe if we skip the drive-thru and toss the soda. But ultra-processed foods are sneakier than a greasy burger. They're the fruit-flavored yogurts, the ready-to-go kale chips, the "high-fiber" breakfast bars wrapped in shiny plastic. Industrially produced and cleverly marketed, they often come loaded with emulsifiers, stabilizers, artificial sweeteners, and other additives that don't exist in nature—or in your grandmother’s kitchen.

We eat more of these than we think. Even when we think we’re eating “healthy.”

And the scariest part? We have no real way to measure how much of this stuff we're actually putting into our bodies. Relying on memory-based dietary surveys is like trying to assess someone’s sleep habits by asking how tired they felt last Tuesday. It’s flawed. We forget. We fudge. We tell ourselves comforting lies.

But your body doesn’t lie.

That’s why a new study from the U.S. National Institutes of Health (NIH) has caused quite a stir. Instead of asking people what they eat, researchers looked at something far more telling: what their bodies say they ate.

More specifically, they measured metabolites—tiny chemical leftovers that remain in your blood and urine after your body digests food. These molecules are the receipts your body keeps. They don’t rely on your memory. They don’t care what you think you ate. They just tell the truth.

In a groundbreaking study published in PLOS Medicine in May 2025, NIH scientists developed a multi-metabolite score—a kind of biochemical fingerprint—to objectively estimate how much energy someone derives from ultra-processed foods. This score could one day replace, or at least enhance, the notoriously unreliable self-reporting tools that have long plagued nutrition research.

And the science behind it is fascinating.

Let’s take Jose, a retired construction worker living alone. He used to eat home-cooked meals made from scratch—beans, rice, fresh chicken. But now he relies mostly on frozen dinners and convenience-store sandwiches. When asked about his diet, he describes it as “simple and mostly healthy.” But his blood doesn’t agree. His metabolite score reveals that roughly 70% of his energy intake comes from ultra-processed foods. He didn’t lie—he just didn’t know.

Or Mia, a single mother working two jobs. She prides herself on feeding her kids breakfast every morning—whole grain cereal with milk, yogurt with fruit, “natural” snack packs. She thinks she’s doing everything right. But those cleverly marketed foods are heavily processed. Her kids are consuming more artificial additives than she realizes. Again, not her fault. But it matters.

The researchers behind the study knew this. That’s why they took a multi-layered approach: part observational, part experimental. First, they gathered dietary and biological data from 718 older American adults who participated in a year-long study. Then they conducted a small but intensive clinical trial involving 20 volunteers, each of whom spent two weeks on a diet composed of 80% ultra-processed foods and another two weeks on a diet with none.

Blood and urine samples were taken throughout, and the data painted a clear picture. Hundreds of metabolites were associated with ultra-processed food consumption. Using machine learning, the team identified patterns in these molecules and created two distinct multi-metabolite scores—one for blood, one for urine—that could reliably detect how much ultra-processed food someone had eaten.

And it worked. The scores accurately distinguished between the “junk food” phase and the “clean eating” phase of the trial. No guesswork. No questionnaires. Just cold, hard chemistry.

It’s the nutritional equivalent of checking your car’s emissions instead of asking the driver how far they think they drove.

To be clear, this tool isn’t perfect yet. The study focused on older adults in the U.S., and eating habits differ across age, culture, and geography. Researchers emphasize the need to refine the score for use in younger populations, ethnic minorities, and communities with different food environments. Still, this marks a major step forward.

Because here’s the thing: nutrition science has always been tricky. Food is deeply personal, cultural, emotional. People don’t eat ingredients—they eat meals, stories, memories. They eat what's affordable and convenient. And in the real world, nobody weighs out their chickpeas or counts every preservative.

But the consequences are real. Diets high in ultra-processed foods have been consistently linked to obesity, type 2 diabetes, cardiovascular disease, and even certain cancers. If we want to understand these connections better—and help people like Erin, Jose, and Mia—we need tools that measure reality, not recollection.

Imagine a future where your yearly check-up includes a metabolite score alongside your cholesterol and glucose levels. Where a simple urine test could tell your doctor how much of your diet comes from synthetic, factory-derived foods. Where public health advice is based not just on surveys, but on the biochemistry of real people.

We’re not there yet. But this study suggests we’re on the way.

As a health professional, I see the value in meeting people where they are. Not everyone has time to cook lentils from scratch. Not everyone can afford fresh salmon or organic kale. But everyone deserves honest feedback from their body. Everyone deserves tools that help them understand what’s going on beneath the surface.

Next time you stand in the supermarket aisle, reading labels that promise “natural,” “low fat,” or “protein-packed,” ask yourself: what will my blood say?

Because the truth is, you can outsmart marketing, you can misreport a food log—but you can’t fake a metabolite.

Your body keeps the score. And finally, we’re learning how to read it.