ClinicaMind - Conversational AI for Healthcare

A comprehensive medical clinic management platform combining AI-powered conversation analysis with a modern web interface for clinical workflows.

Architecture

This is a monorepo containing both frontend and backend components:

• fe/ - React TypeScript frontend with shadcn/ui components
• be/ - Python backend with AI agents for speech processing and medical analysis

Features

Frontend (React + TypeScript)

• Real-time Medical Conversation Analysis - Dynamic UI components that appear synchronized with actual doctor-patient conversations
• Dual Audio Support - Two complete medical encounters (initial visit + 30-day follow-up)
• Comprehensive Medical Components - Physical examinations, assessments, treatment plans, EMG tests, lab results
• Professional Animations - Framer Motion for smooth, polished interactions
• Modern UI Stack - shadcn/ui components with Tailwind CSS and dark mode support

Backend (Python)

• Speech Recognition - Google Cloud Speech-to-Text integration
• Text-to-Speech - Google Cloud TTS for audio generation
• Pain Assessment AI - Machine learning models for pain classification and regression analysis
• Security & Ethics Validation - Automated checks for clinical data handling
• Audio Processing - Real medical conversation analysis with timing synchronization

Getting Started

Frontend Development

cd fe
pnpm install
pnpm dev          # Start development server
pnpm build        # Build for production
pnpm lint         # Run code quality checks
pnpm format       # Format code with Prettier

Backend Development

cd be
pip install -r requirements.txt
python pain_orchestrator.py    # Run main orchestrator
python run_agent_pipeline.py   # Run AI pipeline

Demo Experience

The application showcases a realistic medical conversation analysis system:

1. Patient Case: David S with left upper arm pain
2. Dual Conversations: Initial visit and 30-day follow-up
3. Audio Playback: Press spacebar to start/stop actual medical conversations
4. Dynamic UI: Components appear in real-time as topics are discussed
5. Conversation Switching: Toggle between visits using header buttons

Key Components

• Physical Examination - Range of motion, neurological findings
• Assessment - Chief complaint, pain characteristics, clinical findings
• Treatment Plan - Medications, follow-up instructions
• EMG Testing - Procedures, patient concerns, expected findings
• Lab Results - Color-coded results with clinical interpretation

Technology Stack

Frontend

• React 18 with TypeScript for type safety
• Vite with SWC for fast compilation
• shadcn/ui comprehensive component library
• Tailwind CSS for styling with custom medical theme
• Framer Motion for professional animations
• React Query for server state management
• React Hook Form + Zod for form validation

Backend

• Google Cloud Speech API for speech recognition
• Google Cloud TTS for text-to-speech
• NumPy for numerical computations
• scikit-learn (joblib models) for pain assessment ML
• Pydub for audio processing

Project Structure

ClinicMinds/
├── fe/                          # Frontend application
│   ├── src/
│   │   ├── components/
│   │   │   ├── ui/              # shadcn/ui components
│   │   │   └── medical/         # Medical-specific components
│   │   ├── hooks/               # Custom React hooks
│   │   ├── types/               # TypeScript definitions
│   │   └── App.tsx              # Main demo component
│   ├── public/                  # Static assets including audio files
│   ├── package.json
│   └── CLAUDE.md                # Detailed development guide
└── be/                          # Backend services
    ├── pain_orchestrator.py     # Main orchestration logic
    ├── asr_agent.py             # Speech recognition
    ├── tts_agent.py             # Text-to-speech
    ├── pain_assessment_agent.py # ML pain analysis
    ├── security_ethics_agent.py # Data validation
    ├── *.joblib                 # Trained ML models
    ├── *.m4a                    # Audio conversation files
    └── requirements.txt         # Python dependencies

Key Features

• Real Medical Conversations - Actual doctor-patient audio with synchronized UI
• AI-Powered Analysis - Machine learning for pain assessment and classification
• Professional Interface - Modern, accessible design following medical UI patterns
• Type Safety - Comprehensive TypeScript coverage with strict configuration
• Animation System - Smooth, contextual animations that enhance UX
• Responsive Design - Fixed viewport with independent scroll areas
• Audio Integration - HTML5 Audio API with playback controls

Development Guidelines

• Frontend: Uses pnpm for package management, strict TypeScript, and shadcn/ui patterns
• Backend: Python with Google Cloud APIs and scikit-learn for ML capabilities
• Code Style: ESLint + Prettier for frontend, Python best practices for backend
• Architecture: Component-driven frontend with modular Python backend services

---

For detailed frontend development information, see fe/CLAUDE.md.