ABCD fMRI Social Media and Anxiety Patterns

ISEF Category: Computational Biology and Bioinformatics

Subcategory: Computational Neuroscience  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

Your brain is not a single switch. It is more like a city map with roads that speed up, slow down, and reroute traffic. Resting-state fMRI lets you study those routes while the brain is not doing a task. In the ABCD dataset, you can ask whether social media use lines up with different network patterns in teens who report more anxiety.

What Is It?

This project uses resting-state fMRI, which measures brain activity while a person is lying still and not doing a task. You are not reading thoughts. You are looking at patterns in how different brain regions move together over time. If two regions rise and fall in sync, they may belong to the same network, like two neighborhoods connected by a busy road.

Graph theory turns that brain map into a network. Each brain region becomes a node, and each connection becomes an edge. Then you can measure things like density, clustering, and centrality. Those numbers help you compare groups, such as teens with different levels of social media use or different anxiety scores. The ABCD dataset is useful because it is large, public, and built for studying adolescent brain development.

Why This Is a Good Topic

This is a strong science fair topic because it has real data, clear variables, and room for original analysis. You can test whether brain network measures shift with social media use, anxiety, or both, and you can compare multiple ways of defining the network. The topic connects to adolescent mental health, but you still get to do the computer and statistics work yourself. A strong project here teaches data cleaning, network analysis, and careful thinking about correlation versus causation.

Research Questions

• How does self-reported social media use relate to resting-state network density in ABCD adolescents?
• What is the effect of anxiety score on global efficiency in resting-state fMRI graphs?
• Does the relationship between social media use and graph clustering differ between high-anxiety and low-anxiety cohorts?
• To what extent do centrality measures change across adolescents with different levels of reported social media use?
• Which network metrics best separate teens with higher anxiety scores from teens with lower anxiety scores?
• How does controlling for age, sex, and motion artifact change the association between social media use and brain network topology?

Basic Materials

• Computer with enough memory to handle large CSV and imaging metadata files.
• Stable internet access for downloading public ABCD documentation and data access instructions.
• Spreadsheet software for screening variables and tracking cohorts.
• Python installed with Jupyter Notebook for cleaning data and running network analysis.
• Free text editor or code editor such as VS Code.
• Headphones or a quiet workspace for long analysis sessions.

Advanced Materials

• Access to ABCD raw or preprocessed resting-state fMRI data through the approved data portal.
• High-memory workstation or university server for large-scale preprocessing and graph construction.
• Python neuroimaging libraries such as Nilearn, Nibabel, NetworkX, and Pandas.
• Statistical software or packages for mixed models and multiple-comparison correction.
• ImageJ or FSLeyes for quick visual checks of brain images and masks.
• Atlas files for defining brain regions and network nodes.

Software & Tools

• Python: Cleans ABCD tables, builds connectivity matrices, and runs graph analysis.
• Jupyter Notebook: Keeps your code, notes, and plots in one place while you iterate.
• NetworkX: Calculates graph metrics such as clustering, degree, and centrality.
• Nilearn: Handles neuroimaging data and helps extract resting-state connectivity features.
• R: Runs statistical tests, group comparisons, and regression models for your final analysis.

Experiment Steps

1. Define the exact cohort rules you will use, including age range, social media measure, and anxiety measure.
2. Choose one brain atlas and one connectivity pipeline so your graph metrics stay comparable across participants.
3. Plan how you will clean the data, especially how you will filter out motion-heavy scans and missing questionnaire responses.
4. Select a small set of graph metrics that match your research question, instead of measuring everything at once.
5. Build a comparison plan that includes covariates such as age, sex, and motion, so your results are not misleading.
6. Decide which plots and statistics will show group differences clearly, then predefine what counts as a meaningful effect.

Common Pitfalls

• Mixing questionnaire timepoints with the wrong scan session, which breaks the link between behavior and brain data.
• Ignoring head motion, which can create fake connectivity patterns in resting-state fMRI.
• Testing too many graph metrics at once, which makes the results hard to interpret and easy to overfit.
• Using different preprocessing choices across groups, which can make network differences look real when they are not.
• Treating correlation as causation, which would overstate what social media use and anxiety can tell you from this dataset.

What Makes This Competitive

A competitive version of this project uses careful cohort definitions, strong motion control, and a clear statistical plan before the analysis starts. You can also stand out by testing whether one network metric predicts anxiety better than another, or by comparing multiple brain atlases. Strong interpretation matters too, because you should separate brain connectivity findings from simple behavior correlations. If you explain limits well and show clean, reproducible code, your project looks much more serious.

Project Variations

• Compare positive and negative connectivity graphs to see whether anxiety relates more strongly to one type of edge pattern.
• Repeat the analysis with a different atlas to test whether the result depends on how you split the brain into nodes.
• Add sleep, screen time, or family environment variables to test whether they explain part of the social media and anxiety relationship.

Learn More

• ABCD Study Data Dictionary: Search the NIH ABCD Study site for variable definitions, cohort details, and data release notes.
• NIH ABCD Study: Find public documentation, protocol summaries, and data access steps on the NIH ABCD website.
• NIH PubMed: Search review articles on resting-state fMRI, adolescent anxiety, and graph theory methods.
• NIH Open Access resources: Use NIH-linked method papers and protocol descriptions for neuroimaging analysis ideas.
• MIT OpenCourseWare, Introduction to Algorithms: Use the free course materials to strengthen your thinking about graphs, networks, and computational workflows.
• Network Neuroscience articles: Search this peer-reviewed journal for papers on brain graphs, connectivity, and developmental changes.