Pediatric Fever Chatbot API

Welcome to the official documentation for the Fever Model of Docokids, an AI-driven solution for pediatric fever assessment operating via a conversational API. This API leverages AI to assist in handling pediatric fever-related conversations, integrating seamlessly with platforms like DocoChat.

Features

  • AI-Driven Responses: Provides intelligent responses based on pediatric fever-related queries
  • Advanced Prompt Engineering: Sophisticated conversation system that simulates real pediatric consultations
  • DocoChat Integration: Easily integrates with DocoChat for real-time communication
  • FastAPI Framework: Built using FastAPI for high performance and ease of use
  • Comprehensive Documentation: Detailed guides and references to assist developers
  • Multiple LLM Support: Seamless switching between LLM providers (OpenAI, Gemini, local)

🧠 Prompt Engineering System

The chatbot implements an advanced prompt engineering system that simulates real pediatric consultations through structured conversation phases:

Conversation Phases

  1. INITIAL: First interaction - asks for child’s age
  2. DISCOVERY: Symptom exploration - one question at a time
  3. ASSESSMENT: Detailed evaluation - intensity, duration, behavior
  4. GUIDANCE: Educational guidance and recommendations

Key Features

  • One Question Per Response: Ensures focused, natural conversation flow
  • Context-Aware Questions: Adapts questions based on information already provided
  • Medical Safety Checks: Automatic detection of emergency symptoms
  • Structured Discovery: Progressive information gathering like real medical consultations
  • Pediatric Expertise: Specialized prompts for child health assessment

For detailed documentation on the prompt engineering system, see Prompt Engineering Guide.

Architecture 

Client (WhatsApp bot)
   ↓
FastAPI Application
   ├─ Routers: /conversations, /providers, /health
   ├─ Services: ConversationService, ProviderService
   ├─ Repositories: Redis (fast), Postgres (audit)
   └─ LLM Adapters: OpenAIClient, GeminiClient, LocalLLMClient
  • Implements the Repository and Dependency Injection patterns in FastAPI for scalability and testability.
  • Metrics with Prometheus, JSON logging, and OpenTelemetry traces ensure observability in production.

Tech Stack

  • Python 3.10+
  • FastAPI (ASGI web framework)
  • Pydantic (data validation)
  • SQLAlchemy (async) + PostgreSQL (audit)
  • Redis (aioredis) (conversational state)
  • Prometheus + Grafana (metrics)
  • OpenTelemetry (distributed traces)
  • Sentry (error monitoring)
  • Jekyll (documentation)
  • GitHub Pages (documentation hosting)
  • GitHub Actions (CI/CD)
  • Docker (containerization)
  • Uvicorn (ASGI server)
  • Pytest (testing)
  • Black (code formatting)
  • Flake8 (linting)
  • MyPy (type checking)
  • Alembic (database migrations)
  • JWT (authentication)
  • OpenAPI/Swagger (API documentation)

Getting Started

To begin using the Pediatric Fever Chatbot API, navigate to the Getting Started section.

For detailed information on API endpoints, refer to the API Reference.

Contributing

We welcome contributions! Please see our Contributing Guide for more details.

License

This project is licensed under the GNU GPL v3.0. See the LICENSE file for details.

Support

If you have questions or suggestions, open an issue on GitHub or contact us at agomez@docokids.com.

📄 Additional Documentation Pages

⚙️ Configuring _config.yml

Your _config.yml file should be placed inside the docs/ directory and configured as follows:

```yaml title: Pediatric Fever Chatbot API description: Comprehensive documentation for the AI-powered pediatric fever chatbot. theme: just-the-docs url: https://yourusername.github.io/your-repo-name baseurl: “/your-repo-name”