Protecting FHIR data
summary
A reference DLP architecture for FHIR detects PHI using three concrete steps: discover & classify, de-identify, and re-id risk analysis, running on structured FHIR resources, free text, images and at-rest or real-time streams. De-identification implements character masking, tokenization, bucketing and date-shifting while discovery uses built-in detectors plus custom regex/dictionary patterns. OCR plus pixel replacement handles burned‑in image text. Output keeps referential integrity for analytics and ML while reducing re‑identification risk.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 0:50Reference Architecture
- 1:25DLP
- 4:06Guidelines
- 5:08Example
- 6:25Techniques
- 7:48Use cases
- 8:33Different types of data
- 9:34Data masking
- 12:09Patient Resource Example
- 13:00Unstructured Text Example
- 14:11FHIR Data Architecture
- 15:10Outro
related talks
06mProtecting PHI in Machine Learning Development Environments
01h 59mDeep Dive Workshop – Data Security and Data Protection
01h 12mModelling Privacy Threats of Medical Data and Health Data Spaces
40mData Privacy and Security by Design
39mData Privacy Engineering to power Responsible Enterprise AI
01h 15mDATA PROTECTION ENGINEERING: WHAT IS THE ROAD AHEAD?
46mProtecting sensitive data in huge datasets: Cloud tools you can use
30mProtecting Data While Maintaining Analyzability & Its Business Value: Is It FACT or FICTION