HL7 · FHIR · EMR · Clinical Notes
Every clinical note
is a privacy incident
waiting to happen.
ClinicalIQ™ de-identifies structured and unstructured clinical data - HL7 messages, FHIR resources, free-text clinical notes, and EMR exports - before any of it reaches an AI system, research environment, or third-party vendor.
HL7 + FHIR
Format support
Inline
Per message
Clinical
Entity detection
Zero
Raw PHI shipped
Before · clinical lab message
IdentifierMRN, full name, date of birth, address
Encounterward, bed, attending physician
Orderlab order number, test code
Lab resultwhite blood cell count · haemoglobin
Clinical notehistory, current treatment, referring physician
After ClinicalIQ · de-identified
Identifierpseudonymised
Encounterlocation retained · attending redacted
Orderorder pseudonymised · test code preserved
Lab resultclinical values preserved (unchanged)
Clinical notehistory preserved · physician name redacted
PHI entity detection · free-text clinical note scanned for embedded identifiersClinical values · all lab results fully preservedPseudonymisation · consistent pseudonym across the message set
The challengeStructured fields are easy.
Structured fields are easy.
Clinical notes are where PHI hides.
HL7 segment mapping tools can strip PID fields. What they cannot do is read a clinical note and understand that “patient seen in outpatient clinic with Dr Smith” contains a referring physician's name, or that “the patient's daughter called yesterday” discloses a family relationship.
ClinicalIQ applies intelligent clinical entity detection trained on Australian clinical text to identify and remove PHI from free-text fields across all clinical document types - not just structured segments.
PHI entity detection · clinical note excerpt
Patient Sarah Mitchell, DOB 03/12/1962, presented to Ward 4B with chest pain. Referred by Dr. James Thornton from St Vincent's cardiology. MRN: 4829274. Contact: (02) 9334 2891.
NAMEPatient name, physician name2 found
LOCATIONWard, hospital name2 found
DEMOGRAPHICSDate of birth1 found
IDENTIFIERMedical record number1 found
CONTACT INFOPhone number1 found
Format coverage
Every clinical data format. Covered.
HL7 v2
ORU^R01 (lab results)
ADT^A01 (admissions)
ORM^O01 (orders)
MDM^T02 (documents)
DFT^P03 (charges)
Clinical-domain text de-identification, locale-aware for Australian terminology
FHIR R4
Patient resource
Observation
DiagnosticReport
DocumentReference
AllergyIntolerance
Condition
All resources with PHI-mapped field stripping per profile
Clinical text
Discharge summaries
Progress notes
Referral letters
Radiology reports
Pathology reports
Operative notes
Intelligent detection trained on Australian clinical text patterns
How it works
The ClinicalIQ™ pipeline.
01
Data Ingest
Clinical message or document received via HL7, FHIR API, file drop, or EMR integration
02
Structure Parse
Message parsed. Structured fields mapped to PHI profile. Free-text segments extracted.
03
PHI Detection
Clinical entity detection scans free-text fields. PHI identified and tagged for removal.
04
Policy Check
Release profile applied. Recipient allowlist enforced. Clinical values preserved.
05
Clean Output
De-identified message or document forwarded. Signed audit entry written.
HL7 versions
v2.3 · v2.4 · v2.5 · v2.6 · v2.8 · CDA R2
FHIR versions
R4 (primary) · STU3 (supported) · AU Base profile
De-id model
Intelligent clinical entity detection trained on Australian clinical text
EMR integration
Epic HL7 feeds · Cerner FHIR API · MedicalDirector · Best Practice · Genie
Throughput
Inline processing · FHIR R4 batch async · clinical note processing
Deployment
On-prem agent · HL7 listener · FHIR proxy · REST API
Audit format
Tamper-evident audit format · entity detection log · HREC exportable
Compliance
OAIC APP 11, My Health Records Act, HL7 DS4P, FHIR R4 de-id spec