Early Detect AI

Inspiration

We noticed that many people, especially women undergoing breast cancer screenings, struggle to understand medical reports because of complex terminology, confusing BI-RADS findings, and language barriers. This often creates fear and anxiety before even consulting a doctor. We wanted to build a platform that makes breast health reports and other medical reports easier to understand in a simpler, calmer, and more accessible way using AI.

What it does

EarlyDetect AI is an AI-powered healthcare interpretation platform focused on simplifying breast cancer screening reports and general medical reports. Users can upload reports like mammograms, blood tests, and hormonal profiles to receive simplified explanations through an interactive healthcare dashboard. The platform also provides:

• AI-generated educational insights,
• multilingual voice explanations,
• a contextual healthcare chatbot,
• symptom analysis,
• medical report history,
• and nearby doctor recommendations.

Our goal is to improve healthcare awareness and accessibility, especially for users who may not easily understand clinical reports.

How we built it

We built EarlyDetect AI using React, TypeScript, Tailwind CSS, and Supabase. We used Gemini 2.5 Flash for AI-powered report analysis and chatbot conversations, OCR.space for extracting report text, and Google Cloud Text-to-Speech along with browser speech APIs for multilingual voice explanations. We also created a custom healthcare dashboard system that categorizes findings into areas like:

• CBC Analysis,
• Hormonal Profile,
• Imaging Findings,
• Tumor Marker Analysis,
• and Symptoms Assessment. The platform stores report history, login sessions, and contextual chatbot memory securely using Supabase.

Challenges we ran into

One of our biggest challenges was designing a healthcare experience that felt informative without creating unnecessary fear — especially for breast cancer-related reports. Early versions with percentage-based risk visuals looked too alarming, so we redesigned the dashboard into a more patient-friendly and clinically realistic interface.

We also faced challenges with:

• multilingual voice generation,
• chatbot memory,
• AI consistency across findings,
• OCR reliability,
• and balancing medical seriousness with calm, understandable communication.

Accomplishments that we're proud of

We are proud of building:

• a multilingual AI healthcare assistant,
• a breast cancer report interpretation workflow,
• contextual AI conversations based on uploaded reports,
• multilingual voice explanations,
• persistent report history,
• and a healthcare-focused UI designed to reduce anxiety and improve clarity.

What we learned

Through this project, we learned a lot about:

• healthcare-focused UX,
• ethical AI communication,
• breast cancer screening terminology,
• multilingual accessibility,
• AI integration,
• and designing systems that prioritize trust and emotional safety alongside technology.

What's next for EarlyDetect AI

In the future, we plan to:

• support more healthcare report types,
• improve OCR accuracy,
• integrate real healthcare provider APIs,
• add appointment booking,
• improve multilingual accessibility,
• and collaborate with healthcare professionals to make the platform even more reliable and impactful.

Built With

• biome
• browser-speech-synthesis-api
• gemini-2.5-flash-api
• google-cloud-text-to-speech
• lucide-react
• ocr.space-api
• postgresql
• radix-ui
• react-18
• react-hook-form
• react-router
• recharts
• shadcn/ui
• supabase
• supabase-auth
• supabase-edge-functions
• tailwind-css
• typescript
• vite
• zod