Gene Bridge

• 

  Landing Page

• 

  Upload dataset section

• 

  Lung Cancer Dataset Analysis

• 

  Lung Cancer Dataset AI-Insights and Recommendations

Inspiration

Genomic AI is revolutionizing healthcare, but existing models often fail to provide equitable predictions across diverse populations. We wanted to tackle bias in genomic data analysis by building GeneBridge, a platform that ensures fairness in cancer risk prediction models.

What it does

• The disparities in genomic datasets and their impact on AI models
• How to preprocess and analyze large-scale genomic datasets from NCI GDC
• Techniques for evaluating model fairness and performance metrics

How we built it

• Backend: Python (Flask/FastAPI), Pandas, NumPy, Scikit-learn
• Frontend: Next.js, Tailwind CSS, Recharts
• ML Model: Trained on TCGA-LUAD dataset to predict lung cancer subtypes
• Data Source: NCI Genomic Data Commons (GDC)

Challenges we ran into

• Ensuring dataset quality and avoiding bias in model predictions
• Building an intuitive UI to present fairness metrics effectively
• Integrating AI-driven insights while keeping the platform user-friendly

Accomplishments that we're proud of

•  Successfully built a full-stack genomic AI platform within the hackathon timeframe
•  Integrated ML fairness metrics to ensure equitable predictions across diverse populations
• Processed and analyzed real-world NCI GDC genomic datasets for accurate cancer subtype predictions
• Designed an intuitive dashboard with multiple visualizations for AI-driven insights
• Automated AI-generated recommendations based on fairness and performance metrics

What we learned

• How to process large genomic datasets and extract meaningful insights
• Techniques for evaluating fairness in AI models, particularly in genomic predictions
• The importance of explainability and transparency in AI-driven healthcare applications
• How to rapidly prototype and deploy a scalable AI-powered web app
• Effective collaboration and problem-solving under time constraints

What's next for Gene Bridge

• Expand model training with more diverse datasets to improve accuracy across populations
• Enhance fairness metrics by incorporating additional bias detection techniques
• Improve AI-generated insights with more contextual recommendations for users
• Partner with researchers and healthcare institutions to refine real-world applications
• Extend GeneBridge beyond lung cancer to cover other genomic diseases and risk assessments

Built With

• fastapi
• next.js
• tailwindcss
• xai