Ultra-Processed Food, Sleep, and Metabolic Risk

ISEF Category: Translational Medical Science

Subcategory: Disease Prevention  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

Ultra-processed foods and short sleep both show up in everyday life, but they may also team up in a way that raises metabolic risk. That makes this topic powerful for a science fair project. You can study a real public health question with free data and real statistical tools. You do not need a lab coat to start, just curiosity and careful analysis.

What Is It?

This project studies whether ultra-processed food intake is linked to metabolic syndrome, and whether that link changes when sleep gets shorter. Metabolic syndrome is a cluster of risks, like high waist size, blood pressure, blood sugar, and unhealthy blood fats. Think of it like several warning lights on a dashboard turning on together.

NHANES is a large U.S. health survey that includes diet, sleep, and lab measurements. You can think of it as a giant snapshot of many people at once. Since people are not randomly assigned to eat more ultra-processed food or sleep less, you need causal-inference methods, like inverse probability weighting (IPW) and doubly-robust estimation, to reduce bias from confounding factors such as age, income, activity, and smoking.

Why This Is a Good Topic

This is a strong science fair topic because the question is real, measurable, and tied to daily choices people can actually change. You can test whether diet and sleep work separately, or whether sleep changes the size of the diet effect. That gives you a clear public health story, not just a chart with a trend. You also get to learn data cleaning, regression, and causal inference, which are useful skills for college research.

Research Questions

• How does ultra-processed food intake relate to the odds of metabolic syndrome in NHANES participants?
• What is the effect of short sleep duration on the association between ultra-processed food intake and metabolic syndrome?
• Does the ultra-processed food-metabolic syndrome link differ between people who sleep less than seven hours and those who sleep seven hours or more?
• To what extent do age, sex, income, and physical activity change the estimated effect of ultra-processed food intake on metabolic syndrome?
• Which metabolic syndrome component, waist size, blood pressure, glucose, triglycerides, or HDL, shows the strongest link with ultra-processed food intake?
• How does doubly-robust estimation compare with ordinary logistic regression for this question?

Basic Materials

• Computer with internet access and enough storage for large datasets.
• R or Python installed on your computer.
• NHANES public-use datasets from CDC.
• Codebook and variable documentation from NHANES.
• Spreadsheet software for screening variables and tracking exclusions.
• Statistics notebook for recording model choices and outputs.
• Headphones or a quiet workspace for long analysis sessions.

Advanced Materials

• Computer with internet access and high memory capacity.
• R or Python with survey-analysis packages.
• NHANES linked datasets and documentation from CDC.
• PubMed access for review articles on diet, sleep, and metabolic syndrome.
• Version control system such as Git for tracking code changes.
• A reproducible reporting tool such as R Markdown or Quarto.
• Access to a biostatistics mentor or university lab for method review.

Software & Tools

• R: Runs survey-weighted models, causal-inference workflows, and data visualization for NHANES analyses.
• Python: Supports data cleaning, regression, and reproducible analysis with scientific libraries.
• RStudio: Gives you a simple workspace for writing and testing R code.
• Jupyter Notebook: Helps you combine code, notes, and figures in one file.
• Zotero: Organizes sources, citations, and review articles for your background research.

Experiment Steps

1. Define your outcome, exposure, sleep groups, and covariates before looking at results.
2. Choose the NHANES cycles and variables that match your question, then map every measure to its exact source.
3. Build a clean analysis dataset and decide how you will handle missing values, outliers, and exclusions.
4. Plan the causal model, including the weighting strategy and the main confounders you will adjust for.
5. Set up subgroup and interaction analyses to test whether sleep duration changes the exposure effect.
6. Prepare sensitivity checks that compare your main model with at least one alternative analytic approach.

Common Pitfalls

• Using self-reported diet without checking how NHANES coded ultra-processed food variables, which can break your exposure definition.
• Mixing NHANES cycles without confirming that the survey weights and variable names still match, which can distort estimates.
• Treating metabolic syndrome as one number without building the component definition from the underlying measures.
• Ignoring missing data patterns, which can bias the sample toward healthier or more complete respondents.
• Calling an observed association causal without weighting, adjustment, and sensitivity checks to support that claim.

What Makes This Competitive

A stronger project goes beyond a simple correlation. You can build a careful causal design, justify your confounders, and compare multiple models to see whether the pattern holds. A competitive entry also tests interaction clearly, not just overall association, and reports effect sizes with confidence intervals. If you add sensitivity analyses and show why your result is or is not stable, your project will look much more research-ready.

Project Variations

• Use a different outcome, such as insulin resistance or high blood pressure, to see whether the pattern stays the same.
• Replace sleep duration with sleep regularity or sleep quality, if the NHANES cycle you choose supports that analysis.
• Compare ultra-processed food intake across age groups or sex groups to test whether the exposure effect changes by subgroup.

Learn More

• CDC NHANES: Search the CDC National Health and Nutrition Examination Survey pages for public-use datasets, codebooks, and survey design guidance.
• PubMed: Search for review articles on ultra-processed foods, sleep duration, and metabolic syndrome.
• NIH Office of Dietary Supplements: Use background pages on nutrition-related biomarkers and dietary assessment concepts.
• NCI Dictionary of Cancer Terms: Find plain-language definitions for epidemiology and study-design terms.
• Modern Epidemiology: Use a library copy or preview pages to understand confounding, weighting, and causal inference basics.
• CDC Data Tutorials: Search CDC training materials for guidance on working with complex survey data.