Wearable AFib Screening Cost Model

ISEF Category: Translational Medical Science

Subcategory: Other  ·  Difficulty: Advanced  ·  Setup: Home Setup  ·  Time: 1 to 2 Months

The Hook

A missed AFib case can hide a stroke risk that lasts for years. Your model can test whether a cheap wrist sensor catches enough cases early to justify screening. That turns a public health question into a math project you can run on your laptop.

What Is It?

A Markov model is a way to simulate how people move between health states over time. Think of it like a board game with a few squares, such as healthy, undiagnosed AFib, diagnosed AFib, stroke, and death. Each round, people move from one square to another based on probabilities you choose from public data.

Wearable PPG means a wrist device that reads pulse changes with light. It does not measure heart rhythm as directly as an ECG, but it can flag irregular patterns that may suggest AFib, or atrial fibrillation. Your model asks a simple question with real stakes: if a young adult gets screened with a wearable, does that early signal prevent enough strokes to make the costs worth it?

This kind of project mixes health data, decision making, and simulation. You are not proving the wearable works in a clinic. You are testing how a screening policy might perform when you plug in real-world estimates from CDC data, Medicare claims summaries, and published studies.

Why This Is a Good Topic

This is a strong science fair topic because you can test a real policy question without needing a hospital lab. You can change one assumption at a time, such as screening accuracy, adherence, stroke risk, or device cost, and measure how the outcome shifts. That gives you clear graphs, clear sensitivity analysis, and clear conclusions. It also connects to a public health problem, since AFib can raise stroke risk and screening choices affect both cost and care.

Research Questions

• How does wearable-PPG screening change lifetime stroke risk compared with usual care in young adults?
• What is the effect of screening frequency on the cost per quality-adjusted life year gained?
• Does the model stay cost-effective when wearable false positives increase?
• To what extent does AFib prevalence in young adults change the best screening strategy?
• Which model assumption has the largest effect on net cost, screening uptake, or stroke prevention?
• How does adding confirmatory ECG testing after a positive wearable alert change overall cost-effectiveness?

Basic Materials

• Laptop with R or Python installed.
• Spreadsheet software for organizing parameter values.
• Public CDC disease data or mortality tables.
• Medicare claims summaries or other public cost summaries.
• Published study estimates for AFib prevalence, stroke risk, and screening accuracy.
• Calculator or note-taking app for tracking assumptions.
• Graphing tool such as Excel, Google Sheets, or matplotlib.

Advanced Materials

• Access to R with decision modeling packages.
• Access to Python with pandas, numpy, scipy, and matplotlib.
• PubMed review articles on AFib screening, stroke prevention, and cost-effectiveness.
• Public-use life table data from CDC or NIH sources.
• Published utility weights for health states and stroke outcomes.
• Peer-reviewed estimates for wearable-PPG sensitivity, specificity, and follow-up rates.
• Optional probabilistic sensitivity analysis tools for Monte Carlo simulation.

Software & Tools

• R: Runs the Markov simulation, sensitivity analysis, and cost-effectiveness calculations.
• Python: Builds the model, stores parameters, and makes plots for scenario comparisons.
• Excel or Google Sheets: Organizes input values and helps you check assumptions before coding.
• RStudio: Makes it easier to write, debug, and document your R model.
• PubMed: Helps you find review articles and primary studies for AFib, stroke risk, and screening accuracy.

Experiment Steps

1. Define the health states and the decision you want to compare, such as wearable screening versus usual care.
2. Gather parameter estimates from public sources, then decide which ones come from CDC data, which ones come from Medicare summaries, and which ones come from published studies.
3. Build a base-case model with one clear starting population and one time horizon.
4. Add costs, stroke outcomes, and quality-of-life weights so the model can return a cost-effectiveness result.
5. Test one-way sensitivity analysis so you can see which assumptions change the conclusion most.
6. Compare your base case against alternate scenarios, then check whether your conclusion still holds under realistic uncertainty.

Common Pitfalls

• Using adult screening assumptions without checking whether young-adult risk is much lower, which can make the model overstate benefit.
• Mixing Medicare cost estimates with a young-adult population without explaining why the prices still apply, which can weaken the health economics logic.
• Treating a wearable alert as a diagnosis, which skips the confirmatory ECG step and inflates false-positive harm.
• Choosing too many health states, which makes the model hard to debug and easy to parameterize poorly.
• Forgetting to test sensitivity, which leaves you with one fragile answer instead of a useful range of outcomes.

What Makes This Competitive

A competitive version goes beyond a single base-case result. You can compare several screening pathways, test which assumptions actually drive the answer, and run probabilistic sensitivity analysis so your conclusion includes uncertainty. Strong projects also justify every parameter with public sources and explain why young-adult screening behaves differently from older-population screening. That analytical care makes the project feel like real health policy research, not just a spreadsheet exercise.

Project Variations

• Use different age bands, such as 18 to 24 versus 25 to 39, to see when screening starts to pay off.
• Swap the wearable screen for an app-based pulse irregularity alert and compare the cost-effectiveness gap.
• Add a confirmatory testing step after a positive wearable result and test how that changes false positives, costs, and stroke prevention.

Learn More

• CDC Data and Statistics: Search for public mortality tables, stroke data, and chronic disease summaries that can support model inputs.
• NIH PubMed: Search for review articles on atrial fibrillation screening, wearable PPG accuracy, and cost-effectiveness modeling.
• Medicare claims summaries: Use public claims reports to estimate health care costs and utilization patterns.
• NCBI Bookshelf: Find free textbook chapters on decision analysis, health economics, and Markov models.
• MIT OpenCourseWare: Search for courses in probability, statistics, and modeling that help you build and interpret the simulation.
• Value in Health: Read published cost-effectiveness studies and model structure examples in peer-reviewed health economics.