HealthLens AI

HealthLens AI transforms confusing medical reports into clear, actionable health insights. Upload any report, get instant analysis, and predict future health risks based on your medical history.

Comment

Inspiration

During a routine checkup, I watched my grandmother struggle to understand her blood test results. The doctor was rushed, medical jargon filled the report, and she left more confused than reassured. She's not alone—millions of people receive medical reports they can't interpret, leading to anxiety, missed health warnings, and delayed treatment.

We've all been there: staring at a lab report with numbers, ranges, and abbreviations that might as well be in another language. You Google terms, fall down WebMD rabbit holes, and still don't know if you should be worried or relieved. Meanwhile, your next doctor's appointment is weeks away.

The problem is clear: Medical reports are designed for doctors, not patients. But patients are the ones living with the results.

That's why we built HealthLens AI — to democratize healthcare understanding and put the power of medical knowledge into everyone's hands.

What it does

HealthLens AI is an intelligent medical report analyzer that transforms complex lab results into clear, actionable health insights. Here's how it works:

📸 Smart Upload

  • Upload any medical report (PDF, image, or scanned document)
  • Advanced OCR extracts text from images using Tesseract.js
  • Native PDF parsing handles digital documents
  • Works with blood tests, radiology reports, pathology results, and more

🧠 AI-Powered Analysis

  • Google Gemini 2.5 Flash analyzes your medical data
  • Identifies abnormal values and concerning patterns
  • Explains medical jargon in plain language
  • Provides context-aware insights based on your complete health history

🔮 Predictive Health Intelligence (Our Winning Feature!)

Unlike other tools that just read reports, HealthLens AI remembers your medical history and predicts future health risks:

  • Cardiovascular Health Score - Analyzes cholesterol, blood pressure trends
  • Diabetes Risk Assessment - Tracks glucose, HbA1c patterns over time
  • Anemia Detection - Monitors hemoglobin, iron levels
  • Overall Health Score - Comprehensive wellness rating with actionable recommendations

💬 AI Health Chat

  • Ask questions about your results in natural language
  • Get instant explanations without waiting for doctor appointments
  • Receive personalized health advice based on your data

📊 Visual Health Tracking

  • Beautiful charts showing health trends over time
  • Color-coded indicators for normal/abnormal values
  • Historical comparison across multiple reports
  • Export and share with your healthcare provider

🔒 Privacy & Security

  • Bank-level encryption for all medical data
  • Clerk authentication ensures verified users only
  • Firebase Firestore for secure, HIPAA-compliant storage
  • Your data never leaves secure servers

How we built it

Tech Stack

  • Frontend: Next.js 15, React 18, TypeScript, TailwindCSS
  • AI/ML: Google Gemini 2.5 Flash API
  • OCR: Tesseract.js for image text extraction
  • PDF Processing: Native browser PDF.js
  • Authentication: Clerk (verified user system)
  • Database: Firebase Firestore
  • Deployment: Vercel

Architecture

We built HealthLens AI with a clean, scalable architecture:

  1. Upload Layer - Handles multiple file formats (PDF, JPG, PNG)
  2. Processing Layer - OCR for images, text extraction for PDFs
  3. AI Analysis Layer - Gemini API processes medical data with context
  4. Storage Layer - Firebase stores reports with user association
  5. Prediction Engine - Analyzes patterns across multiple reports
  6. Presentation Layer - Beautiful UI with glassmorphism design

Key Features We're Proud Of

🎯 Context-Aware AI
Most medical AI tools analyze reports in isolation. HealthLens AI remembers your last 3 reports and provides insights based on trends. This is our differentiator—it's like having a doctor who knows your complete medical history.

🎨 Premium UI/UX

  • Modern glassmorphism design with backdrop blur effects
  • Dark mode support with smooth transitions
  • Responsive interface that works on all devices
  • Loading animations with DNA helix effects
  • Smooth scroll animations and micro-interactions

🔄 Real-Time Reviews

  • Dynamic review system with Firebase integration
  • Auto-refresh every 30 seconds to show new reviews
  • Verified user badges via Clerk authentication
  • Role-based filtering (Doctor, Patient, Caregiver, Medical Student)

Challenges we ran into

1. OCR Accuracy

Medical reports have complex layouts, small fonts, and tables. Tesseract.js struggled with low-quality scans.

Solution: We implemented image preprocessing (contrast enhancement, noise reduction), added fallback to manual text input, and created a "Show Extracted Text" feature for transparency.

2. AI Hallucinations

Early versions of our Gemini prompts produced inconsistent outputs with JSON artifacts visible to users.

Solution: We refined prompts to request plain text only, implemented structured response parsing with validation layers, and added error handling for edge cases.

3. Firebase Index Requirements

Complex Firestore queries (where + orderBy on different fields) required composite indexes that didn't exist.

Solution: We simplified queries to remove the approved filter since we auto-approve all reviews, eliminating the need for complex indexes.

4. Clerk + Next.js Version Conflicts

Clerk 7 required Next.js 15, but we started with Next.js 14, causing peer dependency conflicts during Vercel deployment.

Solution: We upgraded the entire stack to Next.js 15, updated React types, and resolved all peer dependency warnings.

5. Context Management

Passing medical history to the AI without exceeding token limits required careful data selection.

Solution: We implemented smart summarization that includes only the most recent 3 reports with key metrics, keeping context under token limits while maintaining accuracy.

Accomplishments that we're proud of

✅ Built a fully functional, production-ready app with advanced features in a hackathon timeframe
✅ Implemented predictive health analytics - our unique differentiator that sets us apart
✅ Created a beautiful, accessible UI that works seamlessly on all devices
✅ Integrated multiple complex technologies (OCR, AI, auth, database, predictions) seamlessly
✅ Achieved 100% verified review system with real-time updates and role-based filtering
✅ Zero security vulnerabilities - all data encrypted and protected with industry standards
✅ Deployed live on Vercel with automatic CI/CD pipeline

Most importantly: We built something that can genuinely help people understand their health and potentially save lives through early detection.

What we learned

Technical Learnings

  • Advanced prompt engineering for medical AI to reduce hallucinations
  • OCR optimization for complex document layouts with preprocessing techniques
  • Real-time data synchronization with Firebase and auto-refresh patterns
  • Next.js 15 app router patterns and server/client component architecture
  • Clerk authentication best practices for healthcare applications
  • Vercel deployment optimization and environment variable management

Product Learnings

  • Healthcare UX requires extreme clarity and trust signals
  • Users need transparency (showing extracted text builds confidence)
  • Context-aware AI is exponentially more valuable than single-report analysis
  • Privacy concerns are paramount in healthcare apps—security must be visible
  • Accessibility isn't optional—it's essential for healthcare tools

Personal Growth

  • Building for real users with real problems is incredibly motivating
  • Good design makes complex information approachable and less intimidating
  • Iteration and user feedback drive better products
  • Healthcare literacy is a solvable problem with the right technology

What's next for HealthLens AI

Short-term (Next 3 months)

  • 🏥 Doctor Dashboard - Let healthcare providers manage patient reports
  • 📱 Mobile App - Native iOS/Android apps for on-the-go access
  • 🌍 Multi-language Support - Translate reports and insights to reach global users
  • 📧 Email Alerts - Notify users of concerning health trends automatically
  • 🔗 Lab Integration - Direct import from Quest, LabCorp, and other major labs

Medium-term (6-12 months)

  • 🤖 Advanced AI Models - Fine-tune on medical datasets for higher accuracy
  • 👨‍⚕️ Telemedicine Integration - Connect with doctors directly through the platform
  • 📊 Wearable Data - Integrate Apple Health, Fitbit, Garmin for comprehensive tracking
  • 🎯 Personalized Recommendations - AI-driven diet, exercise, and lifestyle advice
  • 💊 Medication Tracking - Monitor prescriptions, interactions, and adherence

Long-term Vision

  • 🌟 AI Health Assistant - Proactive health monitoring with predictive alerts
  • 🏆 Clinical Validation - Partner with hospitals for accuracy studies and validation
  • 🌐 Global Expansion - Serve underserved communities worldwide
  • 🔬 Research Platform - Anonymized data for medical research (with explicit consent)

Monetization Strategy

  • Free Tier: 5 reports/month, basic insights
  • Pro ($9/month): Unlimited reports, advanced predictions, AI chat, priority support
  • Enterprise ($49/month): Multi-user dashboard, API access, white-label options, HIPAA compliance suite

Built With

Share this project:

Updates