Researchers tested a new method to figure out how healthy someone’s diet is by combining blood test results with eating habits. Instead of just asking people what they eat, scientists used a computer program to look at blood markers like cholesterol and vitamin levels to predict whether someone follows a healthy diet pattern. In a study of 138 people, this new approach worked pretty well—it correctly identified diet quality about 72-78% of the time. While this is still early research, it suggests that doctors might one day use routine blood work to give people personalized advice about eating better.

The Quick Take

  • What they studied: Can scientists predict whether someone eats a healthy diet by looking at their blood test results instead of just asking them what they eat?
  • Who participated: 138 adults who provided information about their eating habits, had blood tests done, and answered questions about their physical activity and quality of life
  • Key finding: A computer program using blood markers correctly identified healthy eating patterns 72-78% of the time, with blood levels of homocysteine, folate, and vitamin C being the most useful indicators
  • What it means for you: This research suggests doctors might eventually use your routine blood work to give you personalized diet advice, but this method isn’t ready for everyday use yet—it’s still in the testing phase

The Research Details

This was a pilot study, which means it was a small test to see if an idea might work before doing bigger research. The researchers collected information from 138 people one time (called a cross-sectional study). They gathered data in several ways: they asked people about their eating habits using two different methods, took blood samples, measured their height and weight, asked about exercise, and had them fill out quality-of-life surveys. The researchers then used a computer program (machine learning) to look for patterns between what people’s blood tests showed and what their diet quality actually was. They tested four different diet quality measures: the Mediterranean diet, a healthy eating index, the DASH diet, and a vegetarian-focused diet.

Traditional ways of checking what people eat—like asking them to remember everything they ate in a day—aren’t very accurate because people forget or don’t remember correctly. Blood tests are objective measurements that don’t rely on memory. This study explores whether combining blood test results with eating information could give doctors a better, more scientific way to understand someone’s diet quality without relying on what people remember eating.

This is a small pilot study, so the results are preliminary. The computer models explained only 22-35% of the variation in diet quality, meaning other factors beyond blood markers also matter. The accuracy of 72-78% is decent but not perfect. The study was done on one group of people at one point in time, so we don’t know if these results would work the same way for different groups or over time. The researchers were honest that this isn’t ready to be used in real doctor’s offices yet.

What the Results Show

The researchers created computer programs that could predict diet quality using blood test results. The programs worked best when they included information about age, sex, body weight, physical activity, and overall quality of life along with the blood markers. Three blood markers stood out as the most important: homocysteine (a protein building block), folate (a B vitamin), and vitamin C. These markers were significantly linked to how well people followed healthy eating patterns. The computer models were able to correctly identify whether someone was following a high-quality diet about 72-78% of the time, which is better than random guessing but not perfect. When the researchers tested how well the models worked, they got scores between 0.79 and 0.87 on a scale where 1.0 is perfect—this suggests the models had moderate to good ability to predict diet quality.

Other blood markers that showed promise included glucose (blood sugar), triglycerides (a type of fat in blood), and HDL cholesterol (the ‘good’ cholesterol). These markers were associated with diet quality but weren’t as consistently important across all four diet types tested. The study found that different diet quality measures (Mediterranean, healthy eating index, DASH, and vegetarian) had slightly different patterns of blood markers associated with them, suggesting that different healthy eating patterns may have different biochemical signatures in the blood.

This research builds on the growing field of ‘precision nutrition’—the idea that personalized diet advice based on individual characteristics might work better than one-size-fits-all recommendations. Previous research has shown that blood markers reflect eating patterns, but this study is novel in trying to use machine learning to predict overall diet quality. The accuracy rates (72-78%) are reasonable for a pilot study but would need improvement before clinical use. Other studies have tried similar approaches with varying success, and this work adds to the evidence that combining multiple data sources might improve diet assessment.

The study was small (only 138 people), so results might not apply to everyone. The researchers only looked at people one time, so they couldn’t see if the blood markers stayed the same over time or if they changed when people changed their diets. The computer models only explained about 22-35% of why people have different diet qualities, meaning many other factors matter too. The study was done in one location with one group of people, so it’s unclear if the results would work the same for different populations. The researchers were clear this is not ready to replace current diet assessment methods in doctor’s offices.

The Bottom Line

This research is too early to make specific recommendations for patients or doctors. The findings suggest that routine blood work might eventually help doctors understand diet quality better, but more research is needed. If you’re interested in improving your diet quality, traditional methods like talking to a dietitian or using food tracking apps are still your best options right now. (Confidence level: Low—this is preliminary research)

This research is most relevant to doctors, nutritionists, and researchers interested in better ways to assess diet quality. People with chronic diseases like heart disease or diabetes might eventually benefit if this method is developed further. This research doesn’t change what anyone should do about their diet right now. People who are skeptical of blood tests as a way to assess diet quality should know that this method is meant to complement, not replace, traditional diet assessment.

This is very early-stage research. It will likely take 5-10 years of additional studies before this method could potentially be used in regular medical practice, if it proves useful at all. Don’t expect to see this in your doctor’s office anytime soon.

Want to Apply This Research?

  • Track your blood test results (homocysteine, folate, vitamin C, glucose, triglycerides, and HDL cholesterol) every 3-6 months and compare them to your diet quality scores from food tracking to see if patterns emerge in your personal data
  • Use the app to log meals and note when you get blood work done, then review whether your blood markers improve when you follow higher-quality diet patterns for 8-12 weeks
  • Create a dashboard that shows your diet quality score alongside your most recent blood test results to help visualize the connection between what you eat and your biochemical markers over time

This research is preliminary and exploratory in nature. The methods described are not currently recommended for clinical use or personal diet assessment. Blood test results should always be interpreted by a qualified healthcare provider in the context of your complete medical history. If you have concerns about your diet quality or health, please consult with a registered dietitian or your physician. This study does not replace traditional dietary assessment methods or professional medical advice.