Outdoor Exercise Asthma Risk Scores

ISEF Category: Earth and Environmental Sciences

Subcategory: Other  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

Air can look clean and still make breathing harder. For someone with asthma, a jog around the block can feel different from one day to the next. That difference depends on pollution, weather, and the person, not just on the route. Your project can turn those hidden pieces into a risk score.

What Is It?

This idea combines air quality data, weather data, and human symptom reports into one prediction system. OpenAQ gives you pollution measurements from monitors. Sentinel-5P gives you satellite data about gases in the atmosphere. Weather adds conditions like temperature, humidity, wind, and pressure, which can change how pollution behaves and how your lungs react.

Think of it like checking the ingredients before you bake. One ingredient, like ozone or fine particles, may matter. But the full recipe depends on how all the parts mix. Your dashboard would estimate when outdoor exercise is more likely to trigger symptoms for someone with asthma, then compare that estimate with real reports from volunteers.

Why This Is a Good Topic

This is a strong science fair topic because you can test a real prediction with public data and user-reported outcomes. You are not just making a map. You are building and checking a model. That gives you room to study which inputs matter most, whether local versus satellite data perform better, and how well the score matches symptoms in real life. It connects to public health, air pollution, and daily exercise decisions, all of which matter to many families.

Research Questions

• How does adding weather data change the accuracy of an outdoor exercise asthma risk score?
• What is the effect of using local OpenAQ measurements versus Sentinel-5P satellite data on symptom prediction?
• Does including humidity and temperature improve the model’s agreement with self-reported asthma symptoms?
• To what extent can fine particulate matter predict next-day exercise discomfort for volunteer users?
• Which combination of pollution and weather variables best separates low-symptom days from high-symptom days?
• How does model performance change when the score is personalized by user rather than built as a single group model?

Basic Materials

• Laptop or desktop computer with internet access.
• Spreadsheet software or Google Sheets.
• Python installed with pandas, numpy, matplotlib, and scikit-learn.
• Access to OpenAQ data.
• Access to Sentinel-5P data through a public portal or download service.
• Public weather data from NOAA or a similar government source.
• Survey form for volunteer symptom reports.
• Digital notebook for data cleaning decisions and model notes.

Advanced Materials

• Laptop or desktop computer with internet access.
• Python with pandas, numpy, scikit-learn, statsmodels, and plotly.
• Jupyter Notebook or JupyterLab.
• OpenAQ API access.
• Sentinel-5P data access through a public Earth observation portal.
• NOAA weather archives or another government weather database.
• Basic GIS software such as QGIS.
• GitHub for version control and reproducible analysis.
• Secure survey platform for collecting volunteer symptom reports.

Software & Tools

• Python: Cleans the data, builds features, and trains the prediction model.
• Jupyter Notebook: Keeps code, plots, and notes together in one place.
• Google Sheets: Helps you inspect records, spot missing values, and track variables.
• QGIS: Lets you compare pollution patterns with location and weather context.
• ImageJ: Not needed for this topic, so you can skip it unless you add a visual analysis angle.

Experiment Steps

1. Define the exact symptom outcome you will predict, such as wheeze, chest tightness, or missed exercise, so your model has one clear target.
2. Choose the location, time range, and user group you will study, then check that each source has enough overlapping data.
3. Decide which variables you will compare first, such as pollution alone, weather alone, and the combined model.
4. Plan how you will align timestamps and locations so the air data match each symptom report correctly.
5. Build a scoring method that turns raw measurements into a single risk value, then decide how you will validate it against survey data.
6. Set up fairness checks, missing-data rules, and a holdout test so your results do not depend on one lucky split.

Common Pitfalls

• Mixing symptom reports with the wrong date or hour, which breaks the link between exposure and outcome.
• Using too few volunteer reports, which makes the risk score look accurate when it really is unstable.
• Comparing pollution sources with different spatial scales as if they measure the same thing, which can blur the signal.
• Ignoring missing weather or satellite records, which can bias the model toward easier-to-measure days.
• Building one score for everyone, which can hide large differences between users with different asthma triggers.

What Makes This Competitive

A stronger project goes beyond a basic dashboard. You can test whether personalized models beat one-size-fits-all scores, compare satellite and ground measurements, and measure how much each variable adds to prediction quality. You can also use a stricter validation method, such as a held-out time period or a separate user group. That turns your project from a nice visualization into a real study of prediction and health risk.

Project Variations

• Use only local PM2.5 and ozone data to test whether ground monitors alone can predict symptom days.
• Compare urban and suburban users to see whether the same air quality score works in both settings.
• Add pollen or wildfire smoke as a separate factor and test whether it improves the risk model.

Learn More

• OpenAQ: Search the open air quality database for monitor data and API documentation.
• NASA Earthdata: Find Sentinel-5P and other satellite atmosphere data through NASA’s Earth science portals.
• NOAA Climate Data Online: Download weather observations and station history from the National Oceanic and Atmospheric Administration.
• PubMed: Search for review articles on asthma, air pollution, and exercise-triggered symptoms.
• NIH National Heart, Lung, and Blood Institute: Read public health background on asthma triggers and symptom management.