configuring-privacy-policies

To configure clinical text de-identification for HIPAA and GDPR, you select and customize bundled OpenMed privacy profiles that align with Safe Harbor or pseudonymization requirements. This applies per-label actions, quasi-identifier controls, and AnonymizerConfig settings to ensure regulatory compliance.

Yes, you can preserve dates, age, ZIP codes, geography, and clinical concepts while masking direct identifiers. By applying a research limited-dataset policy, the configuration allows fine-grained quasi-identifier handling to maintain data utility for cohort analysis.

The best way to generate synthetic surrogates without leaking PHI is to use strict no-leakage privacy profiles. You can register label-specific custom clinical Faker providers and registered surrogate generators to replace identifiers with isolated, realistic synthetic values.

To set up reversible mappings for pseudonymized clinical data, you configure the Anonymizer with specific arbitration modes and consistency settings. This supports reversible transformations that allow re-identification when authorized, while maintaining GDPR-aligned pseudonymization.

Yes, the de-identification process works offline for internal clinical workflows. It selects and customizes OpenMed privacy policies directly on-device, ensuring clinical text is processed without sending data off-device, which is ideal for strict internal privacy requirements.

You should use custom identifier providers instead of bundled profiles when you need label-specific generators for identifiers like MRNs. This allows you to register isolated per-run clinical Faker providers, offering more specialized surrogate generation than standard HIPAA or PIPEDA profiles.