⚙️ Technical Architecture
AutoAuth’s architecture orchestrates multiple Azure services and techniques to seamlessly process requests, retrieve policies, and generate recommendations.
High-Level Overview
- Knowledge Base Construction: Establish a centralized repository of Prior Authorization (PA) policies and guidelines to streamline the decision-making process.
- Unstructured Clinical Data Processing: Extract and structure patient-specific clinical information from raw data sources using advanced Large Language Model (LLM)-based techniques.
- Agentic RAG: Identify the most relevant PA policy for a clinical case using a multi-layered retrieval approach, supported by Azure AI Search and LLM as the formulator and judge, guided by agentic pipelines.
- Claims Processing: Leverage Azure OpenAI to evaluate policies against clinical inputs, cross-reference patient, physician, and clinical details against policy criteria. Classify the Prior Authorization (PA) claim as Approved, Denied, or Needs More Information, providing clear, evidence-based explanations and policy references to support a comprehensive human final determination.
Components
| Component | Role |
|---|---|
| Azure OpenAI | LLMs for reasoning and decision logic |
| Azure Cognitive Search | Hybrid retrieval (semantic + keyword) |
| Document Intelligence | OCR and data extraction |
| Azure Storage | Document storage |
| Azure Bicep Templates | Automated infrastructure deployment |
| Semantic Kernel | Agentic orchestration of retrieval and reasoning |
| Azure AI Studio (LLMOps) | Model evaluation, prompt optimization, and performance logging |
This integrated design enables a dynamic, AI-driven PA process that is scalable, auditable, and ready for continuous improvement.