Model-Driven Engineering for FHIR Testing: Experiences from Australia's Sparked Accelerator

Fast Healthcare Interoperability Resources (FHIR) is a rapidly emerging and inherently model-driven technology in medical informatics. Its complexity makes it a prime candidate for Model-Driven Engineering (MDE). This paper reports on our experience applying progressively richer MDE techniques to FHIR testing within Australia’s Sparked FHIR Accelerator program. We trace a growth story through three stages: an initial FHIR Mapping Language (FML) tool that solved structural transformation from tabular inputs but proved verbose and hard to maintain; a second-generation Epsilon-based approach with a custom FHIR EMC connector that resolved maintainability through model abstraction but produces structurally rigid test data; and an emerging hybrid direction combining MDE with Large Language Models to address the remaining variability gap. Each stage solved the preceding limitation while revealing a new one. We detail the Epsilon EMC connector, model-driven test data generation for AU Core, and stateful protocol testing for AU eRequesting using generated FHIR TestScript resources executed by FHIR Frog, an open-source TestScript engine. The approach has been deployed at two national connectathons and validated across 17 AU Core resource types and the complete eRequesting lifecycle.