An explainable, low-cost LLM/GraphRAG system for UK NICE Clinical Knowledge Summary (CKS) on Hypertension.

🎯 Current Capabilities

Enhanced Data Pipeline: NICE Guidelines → Advanced Medical Knowledge Graph → Hybrid Retrieval → QA Chain

Clinical Question Answering: GPT-4o-mini powered QA with medical safety prompting

Hybrid Retrieval: Graph-first approach with intelligent vector search fallback

Source Attribution: Complete provenance tracking with confidence scoring

Performance Monitoring: Track retrieval latency, costs, and usage patterns

Cost Tracking: Token-based estimation for OpenAI API usage

Interactive Graph Visualization: Neo4j-style network explorer with color-coded entities and relationships

Clinical Decision Trees: Age-based, ethnicity-based, and conditional treatment pathways

Graph Storage: Persistent medical knowledge graphs in MongoDB

Management Scripts: Comprehensive suite for graph building, data management, and analysis