Agentic Site Reliability Engineering for Self-Healing FHIR Workflow Engines
Healthcare FHIR workflow engines have complicated operational requirements, including high-volume, time-sensitive transactions and complete adherence to the HL7 FHIR Release 4 (R4) standard. Traditional Site Reliability Engineering (SRE) approaches rely on reactive runbooks and manual triage, which are inadequate for the speed and specificity of FHIR fault resolution. This article proposes the FHIR Agentic SRE for Engine Reliability (FASER) conceptual framework, a proposed architectural model for agentic SRE in FHIR workflow engines. The framework introduces a FHIR-specific fault taxonomy covering resource validation errors, transaction rollback failures, subscription notification failures, and consent enforcement errors. FASER employs a four-stage closed-loop operational cycle—Observe, Reason, Act, and Verify—grounded in FHIR-native telemetry and bounded-autonomy remediation. The Reason layer applies LLM-based fault classification with a confidence gating mechanism to constrain autonomous action and ensure human escalation for uncertain or high-risk fault classes. The Verify stage closes the reliability loop by confirming that remediation actions have successfully restored the expected operational state. This article presents the design and analytical justification of the FASER framework. Implementation and empirical validation are identified as future work.
Authors
- Ankita Banerjee
Publication Details
- Journal
- Journal of Computer Science and Technology Studies
- Published
- 2026-11-16
- DOI
- https://doi.org/10.32996/jcsts.2026.8.17
- Primary Topic
- Software System Performance and Reliability
- Type
- article
- Field-Weighted Citation Impact
- 0.00