| Sumario: | Introduction: Creating interoperable clinical data models in FHIR is essential but often labor-intensive. This study explores the use of Generative AI to assist in documenting and structuring FHIR-based data transformation workflows, focusing on TNM cancer staging extraction. Methods: Using FHIR Release 4 and the Plan Definition resource, we modeled transformation processes defined through Business Process Model and Notation (BPMN). The large language model Claude Code (Sonnet 4.5) generated FHIR artifacts from BPMN inputs, which were validated through syntax checks, Implementation Guide compilation, and expert review. Results: AI-assisted generation produced a validated Plan Definition with seven structured activities, accurately representing the TNM extraction workflow. Artifacts were interoperable and passed all FHIR conformance tests after minor revisions. Conclusion / Discussion: Generative AI can effectively support FHIR workflow modeling, reducing complexity and development time. However, expert validation remains essential to ensure semantic accuracy and reproducibility.
|
|---|