Leveraging LLMs for Medicine-Related Questions

Session Track: AI Engineering

Session Time: 05:15 am - 05:30 am November 6

Session description

Nyoman Richardo and Muhammad Hudzaifah Abdurrasyid will present a system that leverages Neo4j and LLMs to answer medicine-related questions using a semantic graph-based approach. The session will explore how they built a structured knowledge graph from medicine data, modeled relationships such as compositions and effects, and connected this graph to a natural language interface using Langchain. You will learn how graph databases can enhance domain-specific question answering, how to construct a medical knowledge graph in Neo4j, and how to integrate it with LLMs to provide more meaningful answers beyond keyword matching. The talk will also cover practical lessons from their implementation, including graph design and prompt engineering for LLMs. This session will be especially useful if you're curious about combining Neo4j with GenAI tools to build accessible and socially impactful applications in the healthcare domain.