BioScan AI

Inspiration

We were inspired by the urgent need for earlier cancer detection. Today, over half of all cancer cases are diagnosed at late stages, drastically reducing survival rates. We asked: What if we could catch cancer before symptoms even appear—using something as simple as a blood test? Advances in liquid biopsy and AI offered a powerful combination, and we saw an opportunity to create a tool that could truly save lives.

What it does

BioScan AI is a prototype platform that simulates early cancer detection using liquid biopsy data and AI. The concept is simple: users upload blood test data, and our system analyzes DNA fragmentation patterns to detect potential cancer signatures and predict the tissue of origin. While the current version doesn’t perform real medical analysis, it demonstrates how the system would work—bringing together data input, deep learning simulation, and user feedback in a single flow.

How we built it

We built the prototype using:

Frontend: React.js for a clean and interactive user interface
Backend: Python with Flask to simulate data processing and AI model calls
AI Simulation: Predefined datasets and model outputs to demonstrate detection logic
Design: Figma and Tailwind CSS for a professional and accessible UX Though the AI functionality is currently simulated, we structured the codebase so it can easily integrate a real model once trained and validated.

Challenges we ran into

Medical data access: Real, labeled ctDNA datasets are not publicly available due to privacy and ethical constraints.
Model limitations: Training an accurate AI model for cancer detection requires massive clinical datasets and resources beyond a hackathon.
Balancing ambition with realism: We had to pivot from building a functioning detection system to showcasing a convincing and educational prototype.

Accomplishments that we're proud of

We created a clean, interactive interface that tells a compelling story of how BioScan AI could work.
We simulated a user journey from blood draw to cancer detection in a way that’s intuitive and medically inspired.
We bridged the gap between cutting-edge research and an accessible web platform.
We turned an ambitious medical AI concept into a working prototype in just 36 hours.

What we learned

How to design for trust and credibility in healthcare applications
The importance of user experience in making complex medical ideas feel simple and usable
How to rapidly prototype data science products even when the full AI model is not ready
That innovation often means balancing scientific potential with practical execution

What's next for BioScan AI

Data partnerships: We're looking into ways to collaborate with researchers or open-source ctDNA datasets to train an actual detection model.
Model development: Building a real deep learning pipeline for ctDNA pattern recognition and tissue-of-origin prediction.
Clinical advisory: Seeking guidance from oncologists and bioinformatics experts to ensure scientific validity.
Regulatory awareness: Starting research into what it would take to turn this into a real, FDA-compliant product.

For now, BioScan AI is a visionary prototype—but with the right data and development, it could one day transform cancer detection worldwide.