auditing-deidentification-runs

Generate signed, PHI-free audit reports for OpenMed de-identification runs.

5.0k|615|Updated Oct 4, 2025
One-click install
npx skills add https://github.com/maziyarpanahi/openmed --skill auditing-deidentification-runs
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: auditing-deidentification-runs
Source: https://github.com/maziyarpanahi/openmed/tree/main/skills/auditing-deidentification-runs
Command: npx skills add https://github.com/maziyarpanahi/openmed --skill auditing-deidentification-runs

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill creates reproducible, tamper-evident audit evidence for clinical text de-identification without retaining or exposing plaintext protected health information.

Core Features & Use Cases

  • PHI-Free Audit Reports: Record redaction spans, labels, confidence thresholds, detector provenance, hashes, and residual risk without storing identifier values.
  • Signing and Verification: Sign reports with HMAC-SHA256 and verify their integrity and optional bindings to the original and de-identified text.
  • Compliance Evidence: Support HIPAA and GDPR reviews, internal quality checks, reproducibility investigations, and defensible release workflows.
  • Use Case: After de-identifying a clinical note, generate and sign an audit report that reviewers can inspect and verify without accessing the original patient information.

Quick Start

Ask the OpenMed skill to run de-identification with auditing enabled, sign the resulting report using a securely stored HMAC key, save the PHI-free JSON record, and verify it later.

Frequently Asked Questions about auditing-deidentification-runs

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a tamper-evident audit trail for clinical text de-identification?

To create a tamper-evident audit trail for clinical text de-identification, run audit-enabled de-identification, generate a PHI-free report with redaction spans and detector provenance, and sign it using HMAC-SHA256. This produces verifiable compliance evidence without exposing plaintext identifiers.

What is a PHI-free audit report and how does it support HIPAA compliance?

A PHI-free audit report records redaction spans, labels, confidence thresholds, content hashes, and residual risk without storing protected health information values. It supports HIPAA compliance by allowing reviewers to verify de-identification decisions and redaction integrity without accessing original patient data.

How do I verify the integrity of a de-identification audit report?

To verify the integrity of a de-identification audit report, use the securely stored HMAC key to check the HMAC-SHA256 signature. This verification validates the report's integrity and its optional bindings to the original and de-identified text, ensuring the audit evidence has not been altered.

Does GDPR require PHI-free audit evidence for clinical text redaction workflows?

GDPR compliance benefits from PHI-free audit evidence for clinical text redaction workflows by providing reproducible, signed records of de-identification decisions. This allows internal quality checks and regulatory reviews to verify redactions and residual risk without retaining or exposing plaintext identifiers.

What is included in an OpenMed de-identification audit record?

An OpenMed de-identification audit record includes redaction spans, labels, confidence thresholds, detector provenance, content hashes, and residual-risk reporting. It captures the full context of redaction decisions while remaining free of plaintext protected health information.

Can I check de-identification residual risk without accessing original clinical notes?

Yes, you can check de-identification residual risk without accessing original clinical notes by reviewing the generated PHI-free audit report. It includes residual-risk reporting and structured span provenance, enabling reproducibility checks and internal review without exposing plaintext identifiers.