D003 敏感信息识别与去标识化

在本地识别并处理患者身份与其他敏感信息，保存去标识化结果和检查证据，为后续临床数据加工提供受控输入。

个人 Skill知识加工12 个文件浏览次数 0

获取完整包获取 SKILL.md

开始使用

下载前阅读 SKILL.md 中的使用条件，并确认所用工具支持对应能力。

输入资料: Required: local source package path Recommended handoff artifacts: D001 source_manifest.json D002 schema_linkage_profile.json Example:D003 may run without D002, but identifier-role inference becomes less reliable and the… 阅读本节完整说明
交付内容: Always produce: phi_gate_report.json phi_gate_report.md deidentified_package/ for supported sanitized artifacts deidentified_manifest.json — source→sanitized artifact/status handoff privacy_review_queue.csv — review-requ… 阅读本节完整说明

获取完整包，保留配套文件的目录结构。
按照原文准备所需工具、依赖与输入资料。
阅读对应使用步骤与输入输出说明，在自己的环境中验证结果。

使用提示摘自原文，配套文件保留来源版本；实际运行条件以各包说明为准。

正文

获取原文件

references/phi-policy.md

Linkage roles

Pseudonymize values in fields identified as patient, patient-or-encounter, encounter, specimen, order, and selected generic identifiers needed for joins.

Direct identifiers

Redact field values for names, phone, email, address, national identifiers, insurance/member/account identifiers and similar direct identifiers.

Birth date

Prefer year-only when a valid year can be preserved without inventing a date. Never synthesize January 1 as a fake full birth date.

Clinical dates

Preserve event dates in v0.1 to maintain timeline semantics. Do not treat the resulting package as Safe Harbor or externally releasable.

Free text

Use known structured identifiers plus deterministic high-confidence patterns. Mark the file as free-text-review-required even if no residual patterns remain.