IKP Asystent Kardiologiczny

AI-powered cardiology pre-screening assistant for the Polish healthcare system. Conducts structured medical interviews and generates comprehensive triage reports for cardiologists.

Live Demo | GitHub

Features

• Two-phase interview flow: Predefined cardiology questions followed by AI-generated follow-ups
• Multi-provider LLM support: OpenAI (including ChatGPT-5), Anthropic Claude, Azure OpenAI
• Professional reports: On-screen display with PDF export
• Polish language: All content in Polish, following gov.pl design system
• Configurable prompts: Edit AI prompts in prompts.json without code changes
• Patient demographics: Collects sex and age for risk-adjusted analysis

Quick Start

# Install dependencies
npm install

# Start development server
npm run dev

Open http://localhost:3000 in your browser.

Configuration

LLM Setup (Required)

Click the settings icon in the header to configure:

1. Select provider (OpenAI, Anthropic, or Azure)
2. Enter your API key
3. Choose model (ChatGPT-5, GPT-4o, Claude Sonnet 4, etc.)
4. Test connection and save

No environment variables required - settings are stored in browser localStorage.

Environment Variables (Optional)

# .env.local

# Supabase (optional)
NEXT_PUBLIC_SUPABASE_URL=...
NEXT_PUBLIC_SUPABASE_ANON_KEY=...

# Default LLM fallback
LLM_PROVIDER=openai
LLM_API_KEY=sk-...
LLM_MODEL=gpt-4o

Usage Flow

1. Enter patient data: ID, sex, age
2. Phase 1: Answer 30+ predefined cardiology questions
3. AI Analysis: System generates personalized follow-up questions
4. Phase 2: Answer 3-8 follow-up questions
5. Report: View and download PDF report for cardiologist

Project Structure

src/
├── app/
│   ├── page.tsx              # Landing page
│   ├── interview/page.tsx    # Interview flow
│   ├── report/page.tsx       # Report display
│   └── api/                  # API routes
├── components/ui/            # UI components
├── lib/
│   ├── prompts.json          # All AI prompts (editable)
│   ├── questions/            # Predefined questions
│   ├── llm/                  # LLM provider abstraction
│   └── store.ts              # Zustand state
└── types/                    # TypeScript types

Customizing Prompts

Edit src/lib/prompts.json to modify AI behavior:

{
  "analysis": {
    "system": "...",  // Cardiologist analysis prompt
    "userTemplate": "..."
  },
  "report": {
    "system": "...",  // Report generation prompt
    "userTemplate": "..."
  }
}

Medical Grounding

Based on:

• ESC Guidelines (European Society of Cardiology)
• ASCVD/EZ-CVD Risk Calculators
• Polish medical interview standards (wywiad lekarski)

Tech Stack

• Framework: Next.js 16 (App Router)
• Styling: Tailwind CSS (gov.pl design system)
• State: Zustand
• PDF: jsPDF
• LLM: OpenAI / Anthropic / Azure APIs

Scripts

npm run dev      # Development server
npm run build    # Production build
npm run start    # Start production
npm run lint     # ESLint

License

MIT