Pii Masking Health Phi 200K: European Medical PII Dataset for Privacy

252,437 entries of European personal health and medical information, part of the larger PII-Masking-2M collection by AI4Privacy. The dataset contains 1,686,246 PII annotations across 49 label types, spanning 23 languages and 29 regions. It was last updated on HuggingFace in April 2026.

Use Cases

• Training PII detection models based on the 49 distinct annotation labels.
• Benchmarking de-identification tools on multilingual European health text across 23 languages.
• Analyzing the distribution of sensitive health information types across different European regions.
• Developing synthetic data generation techniques for privacy-preserving medical AI using the annotated PII structure.

Strengths

• Large annotation volume with 1,686,246 PII labels applied to 252,437 entries.
• Broad European coverage with text in 23 languages and from 29 regions.
• Specific focus on health and medical information, a critical domain for privacy.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality and format require manual inspection.
• Data may reflect geographic and linguistic bias inherent to its European source composition.

Provenance

Source

AI4Privacy via HuggingFace.

Collection Method

Likely collected and annotated as part of a privacy-focused project.

Time Range

null

Freshness

Last updated 2026-04-04 16:18:42; freshness should be verified.

Geography

European regions.