Step 2 — FHIR-to-Graph Ingestion (T1D vs T2D)

What this does: loads the Step 1 corpus (dm_corpus.json), and for every patient instantiates PatientNode / ObservationNode / DiagnosisNode / InterventionNode / OutcomeNode connected by the patient-anchor edges (HAS_OBSERVATION, HAS_DIAGNOSIS, RECEIVES_PLAN, RESULTS_IN — time as an edge property) plus the relational edges (SUGGESTS, INDICATES_PLAN, PRODUCED_OUTCOME, MODULATES, CONSTRAINS). Same code path for T1D and T2D — no special-casing.

Corpus file

PatientNode ObservationNode DiagnosisNode InterventionNode OutcomeNode

Export

GraphML opens in Gephi/yEd. The Neo4j CSV pair follows neo4j-admin database import / LOAD CSV column conventions (id:ID, :LABEL, :START_ID, :END_ID, :TYPE) — schema-ready for later, not loaded into a live instance here.

Step 2 — FHIR-to-Graph Ingestion

Export

Inspect a patient graph

Nodes

Edges

Cohort graph summary (JSON)