• 

  Using past data to forecast and prevent future outbreaks

• 

  Stay informed, stay safe: Predicting COVID-19 risk with data-driven insights

• 

  “Predicting H1N1 flu risk to help prevent future outbreaks.

🦠 EPITRENDS - Disease Outbreak Prediction

Inspiration

We were inspired by the global impact of COVID-19 and seasonal H1N1 Flu outbreaks.
The idea is to use historical data and machine learning to help individuals, healthcare workers, and policymakers anticipate and prevent future outbreaks.

What it does

EPITRENDS allows users to:

• Predict the risk of COVID-19 and H1N1 infections using personal data and symptoms.
  
• Compare predictions across three machine learning models: Logistic Regression, Random Forest, and XGBoost.
  
• Visualize model predictions and accuracies on an interactive dashboard.
  
• Gain insights from historical outbreak trends to inform preventive measures.
  

How we built it

• Framework: Built using Streamlit for interactive web-based forms and dashboards.
  
• Data Handling: Used Pandas and NumPy to process historical outbreak datasets.
  
• Machine Learning Models: Trained Logistic Regression, Random Forest, and XGBoost for both COVID-19 and H1N1 prediction.
  
• Visualization: Created interactive charts using Plotly/Altair (or Matplotlib/Seaborn).
  
• Deployment: Streamlit app designed for easy local or cloud deployment.
  

Challenges we ran into

• Handling large model files (.pkl) for GitHub without exceeding repository size limits.
  
• Ensuring the input form is engaging rather than boring while still collecting all necessary information.
  
• Matching user input features with model training data to avoid prediction errors.
  
• Creating a single dashboard that combines predictions and accuracy for multiple models.
  

Accomplishments that we're proud of

• Built a full-stack, interactive predictive app for two major diseases.
  
• Integrated multiple ML models and compared their accuracies in real-time.
  
• Developed a dashboard for visualization and easy interpretation of predictions.
  
• Made a fun and interactive UI that encourages user engagement rather than just filling forms.
  

What we learned

• How to convert machine learning models into interactive web apps using Streamlit.
  
• Best practices for handling large files and virtual environments in GitHub.
  
• How to encode user inputs to match ML model expectations.
  
• The importance of visual storytelling for presenting prediction results.
  

What's next for EPITRENDS

• Integrate real-time data feeds for COVID-19 and H1N1 to improve prediction accuracy.
  
• Expand to other infectious diseases to make a more comprehensive epidemic prediction tool.
  
• Add alerts and preventive suggestions based on prediction risk levels.
  
• Deploy the app to a cloud platform for global accessibility.
  

Built With

• eda
• kaggle
• machine-learning
• python
• scikit-learn
• streamlit