FHIR in the W.H.O.L.E.: Graphing Clinical Data for Smarter AI Systems

Session Track: Knowledge Graphs

Session Time: 08:00 am - 08:30 am November 6

Session description

We present FHIR in the W.H.O.L.E. (Workflows, Hierarchy, Ontology, Linkage, Encoding), a practical framework for reinterpreting FHIR records as graph-based structures, enabling richer insights and better AI-driven clinical systems. We'll walk through five key principles that transform FHIR data into semantically linked nodes: 1. Workflows—How clinical processes and coding pathways can be mapped and optimized 2. Hierarchy—Understanding nested relationships among diseases, drugs, and diagnostics 3. Ontology—Using formal medical ontologies to ground clinical meaning 4. Linkage—Structuring entity-to-entity relationships across patient data 5. Encoding—Translating clinical concepts into standard vocabularies (e.g., SNOMED CT, LOINC, ICD) The talk will highlight real use cases where graph-based FHIR interpretation supports Clinical AI, interoperability, and decision support. This talk will highlight the use of multimodal AI, agentic frameworks, and how all GNNs play a vital role in orchestrating the medical agents. This work is being co-developed in collaboration with Dr. Julia Hitzbleck as part of our ongoing efforts to build intelligent and interpretable Clinical AI agentic systems at TrialBridge (https://trialbridge.org).