Using FHIR Workflows to deliver personalized health experiences
summary
League models campaign graphs as FHIR PlanDefinition resources, uses the apply operation to generate paired Request/Task resources organized into RequestGroups under a CarePlan, and stores actionable steps as ActivityDefinition. Machine learning (collaborative filtering plus A/Bing of constrained plan variants) selects which activity from node choices to surface; branching logic pauses apply until live data arrives. All of this runs on the core FHIR spec with no extensions, and generated workflows can be published as drafts for clinician review.
AI-written summary · about
