Clara

Inspiration

We were inspired by the urgent need to reduce unnecessary ovarian cyst surgeries. Many women undergo invasive procedures that could be avoided with early risk prediction and data-driven treatment recommendations.

What It Does

Clara is an AI-powered platform that predicts the risk level of ovarian cysts, recommends personalized treatment options, and provides actionable insights to support clinical decision-making.

How We Built It

We developed Clara using:

• Streamlit for the user interface
  
• CatBoost for growth forecasting
  
• Logistic Regression for risk assessment
  
• Cosine Similarity for treatment recommendations
  
• Pandas, Plotly, and Seaborn for analytics and visualization
  

Challenges We Ran Into

• Handling missing clinical data: Dealing with incomplete patient records without compromising accuracy.
  
• Avoiding overfitting: Balancing performance while ensuring generalizability across diverse patient cases.
  
• Ensuring medically sound outputs: Aligning predictions with clinical best practices and validating with expert input.
  

What We Learned

• Applying machine learning in healthcare settings
  
• The value of interpretable and trustworthy models
  
• How to build secure, privacy-first health applications
  

What’s Next for Clara

Our current interface is built with Streamlit—an excellent prototyping tool, but limited in flexibility. Next, we’re designing a more intuitive, robust, and scalable interface tailored to clinicians' workflows and patients’ needs.

Built With

• catboost
• csv
• matplotlib
• numpy
• pandas
• pillow
• plotly
• python
• scikit-learn
• seaborn
• streamlit
• xgboost