LLMs improve privacy-utility trade-off for Dutch clinical note de-identification

Researchers have conducted a comparative study on methods for de-identifying Dutch clinical notes to protect patient privacy while allowing for data reuse. The study evaluated traditional methods like differential privacy (DP) and named entity recognition (NER) alongside newer approaches using large language models (LLMs). Findings indicate that DP mechanisms alone significantly reduce data utility, but combining them with LLM-based preprocessing offers a superior balance between privacy and usefulness for clinical text de-identification. AI

影响 New hybrid approaches combining LLMs with differential privacy may improve the utility of de-identified clinical data for research.

排序理由 Academic paper evaluating privacy-preserving techniques for clinical text de-identification.

在 arXiv cs.CL 阅读 →

• Dutch
• GDPR
• HIPAA
• LLMs

• 论文
• 安全

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →