MedLens

Inspiration

Understanding medical reports shouldn't be hard. People struggle with:

Medical jargon that sounds like a foreign language.
Overwhelming test results and diagnoses they can't interpret.
Language barriers preventing proper health understanding.

We built MedLens for people who need clear, simple explanations of their medical reports in their own language - powered by AI, making healthcare accessible to everyone, regardless of medical knowledge or language barriers.

What it does

MedLens transforms complex medical reports into easy-to-understand explanations using AI. Users upload their medical documents, get simplified explanations, and can ask questions in their preferred language.

Key features:

Smart Document Processing - Upload PDFs or images of medical reports.
AI-Powered Simplification - Complex medical terms explained in plain language.
Multi-language Support - Get explanations in English, Spanish, Urdu, French, Arabic, and Hindi.
Interactive Q&A - Ask questions about your medical report in any supported language.
Text-to-Speech - Listen to explanations with natural voice synthesis.
Privacy-Focused - Your medical data stays secure with local processing.

How we built it

Built as a React + FastAPI web application with multiple AI services:

Login

Supabase Auth: Provides secure user login and access control
Supabase Database: Stores only user metadata

OCR Service

Document upload → Tesseract OCR → Text extraction → Structured medical data

Uses local Tesseract for privacy and pdfplumber for PDF processing
Handles both scanned images and native PDF files

AI Service

Medical text → Google Gemini API → Simplified explanations

Context-aware responses that understand medical terminology.
Language-specific prompts for culturally appropriate explanations.

Translation Service

Simplified text → Google Gemini → Multi-language translations

Supports 6 languages with medical context preservation.
Cultural adaptation for health explanations.

Voice Service

Text → ElevenLabs API → Natural speech synthesis

High-quality voice generation for accessibility
Supports 20+ languages

Tech Stack:

Frontend: React with Vite, Tailwind CSS, React Router
Backend: FastAPI with Python
Middle ware: CORS
AI: Google Gemini 2.0 Flash
OCR: Tesseract (local), pdfplumber
Voice: ElevenLabs API
Storage: In-memory state management

Challenges we ran into

Language Barriers - Medical terms don't translate directly.
API Integration - Managing multiple AI services with different response formats.
Privacy Concerns - Balancing AI capabilities with medical data security.
CORS Issues - Frontend-backend communication across different ports.

What we're proud of

Real Document Processing - Actually extracts text from real PDFs instead of dummy data.
Multi-language Chat - Independent language selection for Q&A vs report analysis.
Privacy-First Design - Local OCR processing keeps sensitive data secure.
Seamless Integration - Smooth workflow from upload to AI analysis to voice playback.
Accessibility Features - Text-to-speech and multi-language support for diverse users.

What we learned

Local vs Cloud Processing - OCR works better locally for privacy and reliability.
Language Context Matters - Medical explanations need cultural adaptation, not just translation.
User Experience is Key - Clear visual feedback during processing prevents user confusion.
API Reliability - Multiple fallback options are essential for production apps.

What's next for MedLens

More Languages - Expanding to support 25+ languages including regional dialects.
Advanced AI - Smarter medical explanations with personalized health insights.
Mobile App - React Native version for on-the-go access.
Health Records - Secure storage and management of multiple medical reports.
Insurance Integration - Help users understand coverage and find affordable care options.