Aller au contenu principal

PHI De-identification

Category: Security
Node id: phi-deidentify

Standards: HIPAA, GDPR

Detect and de-identify protected health information (PHI) in clinical text, FHIR resources, or HL7 messages using NLP-based named entity recognition with configurable strategies.

What it does​

  • PHI detection via transformer NER
  • 5 de-identification strategies
  • FHIR/HL7 aware
  • 8 languages
  • Configurable field policies

Runtime​

  • Performance tier: variable

Input ports​

NameTypeRequiredDescription
texttextNoPlain clinical text to de-identify
resourcejsonNoFHIR resource or Bundle to de-identify
hl7_messagetextNoHL7 v2 pipe-delimited message to de-identify

Output ports​

NameTypeRequiredDescription
resultjsonNoDe-identified output (text, FHIR resource, or HL7 message)
logjsonNoAudit trail of all transformations
statsjsonNoEntity counts, timing, model info
successbooleanNoBoolean success indicator
errortextNoError message if de-identification failed

Settings​

Other​

SettingNameTypeRequiredDefaultAllowed valuesDescription
LanguagelanguageenumNoenen, fr, de, it, es, nl, hi, teText language
Model Sizemodel_sizeenumNosmallsmall, largesmall (fast) or large (accurate)
Default Strategydefault_strategyenumNoredactredact, mask, hash, fake, dateshiftDefault de-identification strategy
Entity Policiesentity_policiesjsonNo{}Per-entity-type strategy overrides
Resource Policiesresource_policiesjsonNo[]Per-FHIR-field strategy overrides
SaltsaltstringNoSalt for hash/dateshift determinism
Date Shift Daysdate_shift_daysnumberNo0Date shift offset in days (0 = random)
Confidence Thresholdconfidence_thresholdnumberNo0.7NER confidence threshold (0.0-1.0)
Strict Modestrict_modebooleanNofalseReserved for future residual-PHI blocking support; currently not enforced

See Build any workflow for how to put nodes together and Ports, types and data flow for how port types connect.

Quelque chose n'est pas clair ? Envoyez vos commentaires sur la documentation par courriel.