deidentifying-clinical-text

Remove or transform protected health information from clinical free text.

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill removes, masks, replaces, hashes, or date-shifts patient identifiers in clinical free text so notes can be shared, stored, or analyzed without exposing protected health information.

Core Features & Use Cases

  • Multiple de-identification methods: Choose masking, removal, replacement, hashing, or date shifting based on privacy and research requirements.
  • Safety and privacy controls: Configure confidence thresholds, bundled policies, consistent seeded surrogates, secure mappings, multilingual processing, and audit reports.
  • Clinical workflows: De-identify medical notes, discharge summaries, research datasets, and text prepared for downstream analysis while keeping processing on-device.
  • Related operations: Preview detected spans with extract_pii and restore authorized text with reidentify when a protected mapping is available.

Quick Start

Use OpenMed to de-identify the attached clinical note with masking, a confidence threshold of 0.7, and the hipaa_safe_harbor policy, then review the resulting labels and offsets without displaying raw patient identifiers.

Frequently Asked Questions about deidentifying-clinical-text

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

FAQPage Schema
How do I de-identify clinical text to remove protected health information?

To de-identify clinical text, you can apply masking, removal, replacement, hashing, or date shifting to patient identifiers in medical notes. This process ensures clinical free text can be safely shared or analyzed without exposing protected health information.

Can I use on-device AI to de-identify medical notes for HIPAA Safe Harbor compliance?

On-device AI de-identification of medical notes supports HIPAA Safe Harbor compliance through bundled privacy policies. It processes clinical free text locally, removing or transforming personal identifiers without sending data to external servers.

What is the best way to remove PII and PHI from research datasets locally?

The best way to remove PII and PHI from research datasets locally is using on-device de-identification with configurable confidence thresholds. You can apply consistent seeded surrogates and reversible mappings to maintain data utility while protecting patient privacy.

Does clinical text de-identification support multilingual inputs and audit reporting?

Clinical text de-identification supports multilingual inputs and generates audit reports for compliance tracking. It detects and transforms protected health information across multiple languages while providing detailed records of all de-identification actions taken.

How do I re-identify de-identified text when an authorized mapping is available?

You can re-identify de-identified text using the reidentify operation when a protected mapping is available. This restores the original patient identifiers in clinical notes for authorized users who possess the secure mapping data.

When should I use date shifting versus masking for de-identifying discharge summaries?

Use date shifting for discharge summaries when you need to preserve temporal relationships in clinical data, and use masking when complete removal of date identifiers is required. Both methods transform protected health information while maintaining different levels of data utility.