ONC HTI-5 proposed rule advances FHIR-first interoperability and AI-enabled data exchange

The HTI-5 rule represents ONC’s most direct signal yet that the agency sees FHIR interoperability as foundational infrastructure for AI — not just as a data exchange standard for patient access and provider communication. The explicit reference to AI-enabled interoperability solutions in the rule text is new language that acknowledges AI agents as a category of software that needs reliable, standards-based data access to function effectively.

The removal of C-CDA certification criteria is the structural change with the most operational significance. C-CDA — the Consolidated Clinical Document Architecture — has been the legacy interoperability format that EHRs have supported for over a decade. It is significantly more complex and less machine-readable than FHIR resources. Retiring C-CDA requirements removes a legacy burden but also removes a compatibility layer that some older and community-hospital EHR installations have relied on. The transition period will require careful management.

The 29 new USCDI v7 data elements include several that are particularly relevant to AI applications: more granular social determinants of health data (housing, food security, transportation), standardized disability status coding, and additional laboratory result types that are frequently used in predictive models. Better standardization of these elements reduces the feature engineering work required to deploy AI models across different EHR environments — a meaningful practical benefit.

For health systems planning their AI roadmaps, FHIR R4 implementation completeness is increasingly a prerequisite for advanced AI deployment. Systems that have completed thorough FHIR implementations will have a structural advantage in deploying AI agents that need real-time patient data access.