MedLog Protocol: Interoperability Layers for AI Medical Model Logging
Interoperability - MedLog
Summary
The MedLog protocol defines the information that should be recorded at each invocation of a medical AI model. By leveraging existing open standards such as W3C PROV, FHIR, and OpenTelemetry, it achieves scalable and vendor-agnostic data exchange.
Details
The MedLog protocol defines what information must be recorded every time a medical AI model is invoked. It can be implemented using existing open standards and tooling, providing complementary ways to represent, exchange, and analyze data. Interoperability is achieved through multiple layers: W3C PROV and OpenLineage handle provenance and lineage; USCDI defines high-level health data classes; and FHIR provides structured resource representations. Operational telemetry is supported by OpenTelemetry, while EHR audit logs offer real-world event-level logging data. These combined layers enable scalable, vendor-agnostic deployments, significantly lowering the barrier for multi-institutional analyses. The protocol also provides field mappings linking MedLog fields to USCDI, FHIR, and OpenTelemetry concepts.
Original content copyright by respective publishers