Skip to main content
Version: v0.1.0

Azure OpenAI Integration

Guide to configuring Azure OpenAI for AI-powered features in Ciyex EHR.

Overview​

Ciyex EHR leverages Large Language Models (LLMs) for:

  • Clinical note summarization
  • Smart search capabilities
  • Drafting patient responses (provider review required)
  • Voice-to-text transcription correction

Prerequisites​

  • Azure Subscription
  • Azure OpenAI resource created
  • Models deployed (e.g., gpt-4, gpt-35-turbo, text-embedding-ada-002)

Configuration​

Configure application.yml:

azure:
  openai:
    endpoint: https://your-resource-name.openai.azure.com/
    api-key: ${AZURE_OPENAI_API_KEY}
    deployment:
      chat: gpt-4-deployment
      embeddings: embedding-deployment

Features​

Clinical Summarization​

Automatically generates a summary of the patient's history or recent encounter notes.

public String summarizeEncounter(String encounterText) {
    ChatCompletionsOptions options = new ChatCompletionsOptions(
        Arrays.asList(new ChatRequestSystemMessage("Summarize the following clinical note into SOAP format."),
                      new ChatRequestUserMessage(encounterText))
    );
    // Call Azure OpenAI SDK
}

Security & Privacy​

HIPAA Compliance:

  • Azure OpenAI operates within your Azure compliance boundary.
  • Zero Data Retention: Configure Azure OpenAI to not store prompts or completions for abuse monitoring (requires Microsoft approval) to ensure max privacy.
  • Encryption: Data is encrypted in transit and at rest.

Cost Management​

  • Token Limits: Set max tokens per request.
  • Caching: Common queries are cached to reduce API calls.
  • Rate Limiting: Per-user rate limits to prevent abuse.

Local Testing​

For local development, you can mock the OpenAI service to avoid costs.

# application-dev.yml
azure:
  openai:
    enabled: false # Uses MockOpenAIService