Lehigh University Flu Forecast Dashboard

A Streamlit web application for visualizing flu and influenza-like illness (ILI) forecasts at Lehigh University, developed by the Computational Uncertainty Lab.

Features

🔍 Quick Look

• Current Status Alerts: Visual indicators showing LOW/MEDIUM/HIGH flu and ILI activity compared to historical data
• 4-Week Ahead Forecasts: Probability-based forecasts showing chances of increase/decrease
• Target Selection: Toggle between ILI and Flu Cases
• Temporal Forecasts:
  • Absolute case counts with 95% prediction intervals
  • Percentage positive trends with historical comparisons
• Interactive Visualizations: Altair charts with tooltips and zoom capabilities
• Current Week Indicator: Visual marker showing the most recent data point

📊 Detailed Look

• 4-Week Panel Views: Side-by-side comparison of ILI and Flu forecasts
• Historical Comparisons: Black dots showing historical averages for the same week across seasons
• Prediction Intervals: Blue shaded 80% uncertainty bands
• Median Forecasts: Blue dots indicating the predicted median values
• Weekly Labels: Clear labeling with MMWR week numbers and end dates

📈 Key Metrics

• Current week case counts and percentages
• Historical context for current activity levels
• 4-week horizon forecasts with uncertainty quantification
• Season-specific tracking (currently 2025/26 season)

Setup

1. Install Dependencies:

   cd webapp
   pip install -r requirements.txt

2. Run the Application:

   streamlit run main.py

3. Access the Dashboard: Open your browser to http://localhost:8501

Data Structure

The app expects the following data structure:

webapp/
├── forecasts/
│   └── 2025_26/
│       └── tempo/
│           ├── flu/
│           │   ├── *_tempo_forecast_flu.csv
│           │   └── *_tempo_forecast_prob_above_median_flu.csv
│           └── ili/
│               ├── *_tempo_forecast_ili.csv
│               └── *_tempo_forecast_prob_above_median_ili.csv
├── analysis_data/
│   ├── weekly_data.csv                    # Lab-confirmed flu cases
│   ├── influenza_like_illness.csv         # ILI diagnoses
│   └── from_week_to_season_week.csv       # Week mapping data

File Structure

webapp/
├── main.py                    # Main app with data loading and navigation
├── app.py                     # Single-page version (legacy)
├── requirements.txt           # Python dependencies
├── complogo.001.png          # Lab logo
├── pages/
│   ├── __init__.py
│   ├── quick_look.py         # Main forecast page with alerts and selections
│   └── detailed_look.py      # 4-week panel comparison view
├── forecasts/                # Forecast data organized by season
└── analysis_data/            # Historical and current observation data

Usage

1. Navigate: Use the sidebar dropdown to switch between "Quick Look" and "Detailed Look"
2. Select Target: In Quick Look, toggle between ILI and Flu Cases using the segmented control
3. View Alerts: Check the colored alert boxes for current activity levels
4. Interpret Forecasts:
   • Red/upward: Probability of increase above historical median
   • Blue/downward: Probability of decrease below historical median
5. Explore Data: Hover over charts for detailed tooltips, zoom and pan for closer inspection

Forecast Interpretation

• ILI (Influenza-Like Illness): Diagnosed when fever >38°C with respiratory symptoms
• Flu Cases: Lab-confirmed influenza via testing
• Historical Average: Black dots represent the mean of past observations for that week
• Predicted Median: Blue dots show the forecast median
• 80% PI: Blue shaded region containing 80% of probable outcomes
• 95% PI: Wider interval shown in temporal charts

Development

To add new pages:

1. Create a new file in pages/ directory with a show() function
2. Import the module in main.py
3. Add to the page selection dropdown in the sidebar

Team

Computational Uncertainty Lab, Lehigh University

• PI: Prof. Thomas McAndrew
• Flu-Crew Team: Kelechi Anyanwu, Ava Baker, Ava Delauro, Eric Shapiro, Holden Engelhardt, Lela Boermeester

IRB Approval

This project was determined to be not human subjects research by the Lehigh University IRB (2367498-1) on September 24, 2025.

Resources

• Computational Uncertainty Lab
• Pennsylvania Department of Health Respiratory Dashboard
• CDC Influenza Information
• WHO Influenza Information
• Lehigh University Health and Wellness Center