Pathogen Platformer - Burndown

[zachpeltz]

Backend

• Preprocess NCBI Pathogen data for:
• Genome sequences
• Resistance genes
• Outbreak location metadata
• Implement CNNs for microscopy-based image classification
• Implement Decision Tree & Logistic Regression models to classify:
• Infection severity
• Treatment options
• API endpoint:
• Receives collected genome fragments
• Returns pathogen name, resistance status, and treatment suggestions

Frontend (UI Integration)

• Classic Mario-style UI:
• Side-scrolling 2D platformer
• Collectibles = genome fragments and powerups (lab coats, medkits, etc.)
• At checkpoints:
• Display DNA/microscopy image
• Provide ML prediction popup (e.g., “Predicted: Ebola – High Risk – Use Containment Strategy X”)
• Health bar UI shows player health if infected or slowed by hazards
• Lab Station UI:
• Choose a “treatment card” based on ML output
• Confirm and see impact visually (area becomes decontaminated)

Game Logic & Mechanics

• Randomly spawn pathogens per level based on difficulty and ML model
• Assign threat levels (low, medium, high) to each level based on real-world data
• “Sequencing lab” checkpoint mechanic requires:
• At least 3 genome fragments to activate ML predictions
• Introduce environmental hazards:
• Contaminated water (virus slows player)
• Mutated spores (require mask power-up to pass safely)
• Winning condition:
• Clear infection zones using correct strategies
• Reach the final research center and publish findings

Testing & Optimization

• Unit testing for ML predictions: Match output with actual pathogen labels
• Game difficulty testing: Balance between fun and accuracy (make ML predictions useful but not overpowered)
• Frame rate optimization: Ensure smooth scrolling and loading
• User feedback A/B testing:
• Which ML output display is clearer?
• Does the “lab station” concept break game flow or enhance it?