We’re thrilled to share that our Chief AI Scientist, Michael Banf, will be speaking at Neo4j Developer Conference - Nodes 2024 on November 7th! He’ll dive into how transforming documents into hierarchical graph structures enables precise retrievals—a breakthrough approach with applications across industries.

Retrieval-augmented generation (RAG) has become a powerful technique for increasing the accuracy and reliability of generative AI by linking it with critical information from external sources, such as internal company documents. This method has particularly valuable applications when precision is non-negotiable, especially in areas involving sensitive data. Traditionally, RAG splits documents into sections for semantic comparison with search queries, using this context to guide language models in generating accurate responses.

However, recent advancements show that by including a document’s structure alongside semantic relationships, retrieval precision can be significantly enhanced. For example, relevant content often extends beyond immediate matches, and structural insights—like previous document sections—can bring essential context into view. Graph structures, uniquely suited for representing these relationships, are key in enabling this sophisticated approach.

At Nodes2024, Michael will share recent research demonstrating how we use knowledge graphs built from medical literature to tackle the challenge of automating documentation for doctor-patient consultations. Our AI solution, AdiuHealth, converts medical guidelines into knowledge graphs, enabling accurate consultation summaries enriched with follow-up analyses, such as guideline-based treatment recommendations. This breakthrough allows doctors to reduce documentation time and focus more on patient care and research.

Check out Michael’s session, More Patience for Patients: How LLMs and Graphs Enable Doctors to Focus on What TheySigned Up For, to see how RAG and knowledge graphs are transforming AI applications in healthcare and beyond.