The LLM for Medical Diagnostics - AI-Powered Multimodal Healthcare Platform
MediLens is a production-grade, unified AI diagnostic platform that combines multiple medical specialties into one seamless interface. Healthcare professionals can access specialized, validated diagnostic pipelines for retinal imaging, chest X-rays, ECG analysis, speech biomarkers, dermatology, motor assessment, cognitive testing, and more—all in one unified platform.
| Module | Purpose | Accuracy | Processing |
|---|---|---|---|
| RetinaScan AI | Diabetic retinopathy grading, 12 biomarkers, Grad-CAM heatmaps | 93% | <2s |
| ChestXplorer AI | Pneumonia, TB, COVID-19, lung cancer detection | 97.8% | <2.5s |
| CardioPredict AI | ECG arrhythmia classification, HRV analysis, 15+ biomarkers | 99.8% | <2s |
| SpeechMD AI | Parkinson's, dementia detection via voice biomarkers | 95.2% | <3s |
| SkinSense AI | Melanoma & skin lesion detection (ABCDE criteria) | 94.5% | <2s |
| Motor Assessment | Movement pattern analysis, tremor detection | 93.5% | <2s |
| Cognitive Testing | Memory & executive function assessment | 92.1% | <2s |
| NeuroScan AI | Brain MRI/CT scan analysis | 91.4% | <3s |
| RespiRate AI | Respiratory sound & spirometry analysis | 93.2% | <2s |
| Multi-Modal Fusion | Combined cross-modal analysis | 96.8% | <3s |
| NRI Fusion | Unified neurological risk index (0-100 scale) | 97.2% | <2s |
- Quality Gates - AI rejects degraded inputs before inference
- Uncertainty Quantification - Confidence scores with human-review flagging
- Explainable AI - Grad-CAM heatmaps, biomarker breakdowns, clinical summaries
- AI Orchestrator - LLM-powered cross-modal synthesis
- Voice Explanations - Amazon Polly TTS for spoken clinical summaries
- Real-time Dashboard - WebSocket system health monitoring
- Audit Logging - Cryptographic traceability (HIPAA/FDA ready)
- Dark Mode - Optimized for radiology reading rooms
- Next.js 16 (App Router, React Server Components, TypeScript)
- Tailwind CSS with medical-grade design system
- Framer Motion for 60fps animations
- Clerk for authentication
- Bun package manager & runtime
- FastAPI (Python 3.12+, async microservices)
- Modular Pipeline Architecture (Input -> Preprocessing -> Core -> Explanation -> Output)
- uv for Python dependency management
- Pydantic for strict data validation
- Uvicorn ASGI server
- CLAHE preprocessing for retinal imaging
- Parselmouth/Praat for acoustic feature extraction
- R-peak detection for ECG analysis
- Grad-CAM for visual explanations
- Amazon Polly for voice synthesis
- Amazon Polly - Neural TTS for clinical explanations
- AWS - Cloud-native ready infrastructure
MediLens/
├── frontend/ # Next.js 16 Application
│ ├── src/app/ # App Router pages
│ │ ├── dashboard/ # All diagnostic modules
│ │ ├── about/ # Technical architecture
│ │ └── vision/ # Mission & roadmap
│ └── src/components/ # Reusable UI components
├── backend/ # FastAPI + ML Pipelines
│ └── app/pipelines/ # Modular diagnostic engines
│ ├── retinal/ # RetinaScan AI (v4.0)
│ ├── speech/ # SpeechMD AI (v3.0)
│ ├── cardiology/ # CardioPredict AI
│ ├── radiology/ # ChestXplorer AI
│ ├── voice/ # Amazon Polly TTS
│ ├── explain/ # AI Orchestrator
│ └── chatbot/ # LLM integration
├── docs/ # Technical documentation
└── data/ # Demo assets
- Node.js 24+ or Bun 1.0+
- Python 3.12+
- Git
cd frontend
bun install
bun run dev
# -> http://localhost:3000cd backend
uv venv
uv pip install -r requirements.txt
uvicorn app.main:app --reload
# -> http://localhost:8000| Method | Endpoint | Description |
|---|---|---|
| POST | /api/retinal/analyze |
Retinal fundus analysis |
| POST | /api/speech/analyze |
Voice biomarker analysis |
| POST | /api/cardiology/analyze |
ECG signal analysis |
| POST | /api/radiology/analyze |
Chest X-ray analysis |
| GET | /api/{pipeline}/health |
Pipeline health check |
| GET | /api/{pipeline}/info |
Pipeline metadata |
| Method | Endpoint | Description |
|---|---|---|
| POST | /api/voice/speak |
Text-to-speech (Amazon Polly) |
| POST | /api/voice/generate |
Generate voice for AI explanations |
| POST | /api/explain/analyze |
LLM-powered result explanation |
| GET | /api/voice/voices |
List available voices |
| Method | Endpoint | Description |
|---|---|---|
| GET | /health |
Backend health check |
| GET | /docs |
Swagger API documentation |
- HIPAA Compliance - Encrypted data, audit logging, access controls
- WCAG 2.1 AA - Full accessibility compliance
- ETDRS Standards - Retinal grading compliance
- Performance - <3s processing for all pipelines
- Export - PDF reports, clinical summaries
bun install # Install dependencies
bun run dev # Development server
bun run build # Production build
bun run lint # ESLint validationuv venv # Create virtual environment
uv pip install -r requirements.txt # Install dependencies
uvicorn app.main:app --reload # Development server
python scripts/verify_all.py # Verify all pipelinesNEXT_PUBLIC_API_URL=http://localhost:8000
NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY=pk_...
CLERK_SECRET_KEY=sk_...AWS_ACCESS_KEY_ID=...
AWS_SECRET_ACCESS_KEY=...
AWS_DEFAULT_REGION=us-east-1
OPENAI_API_KEY=sk-...git clone https://github.com/steeltroops-ai/MediLens.git
cd MediLens
git checkout -b feature/your-feature
# Make changes
git commit -m "Add feature"
git push origin feature/your-featureProprietary. All rights reserved.
Email: steeltroops.ai@gmail.com Project: MediLens - The LLM for Medical Diagnostics