Clinical Text De-identification: Protecting PHI from FHIR/HL7 Data

Details

This service addresses both structured data (like HL7 messages and FHIR resources) and unstructured text, particularly focusing on free-text clinical notes where PHI often hides. Unlike simple field stripping, ClinicalIQ uses locale-aware entity detection—specifically for Australian terminology—to identify and remove contextual identifiers from narrative text. This includes names, addresses, dates of birth, and medical record numbers, even when embedded in sentences (e.g., identifying a referring physician's name). The platform supports multiple formats, including HL7 v2 through FHIR R4, covering key resources like Patient, Observation, and DiagnosticReport. The five-step pipeline ensures that while PHI is pseudonymized or removed, critical clinical values (like lab results) are fully preserved. This capability helps organizations meet strict compliance requirements such as OAIC APP 11 and the My Health Records Act, making it valuable for developing AI systems using sensitive patient data.