Differentially Private De-identification of Dutch Clinical Notes: A Comparative Evaluation

2026-04-23

AI Summary

This study addresses the challenge of balancing privacy preservation and data utility in Dutch clinical texts under regulatory frameworks such as GDPR and HIPAA. It presents the first systematic comparison of differential privacy (DP), named entity recognition (NER), and large language models (LLMs) for de-identifying Dutch clinical notes, and proposes a novel hybrid strategy that combines LLM-based preprocessing with DP. Experimental results demonstrate that applying DP alone substantially degrades data utility, whereas integrating LLM or NER preprocessing significantly improves the privacy–utility trade-off. Extrinsic evaluations—including entity and relation classification—confirm that the proposed hybrid approach effectively maintains downstream usability while providing strong privacy guarantees.

Technology Category

• Machine Learning: Privacy
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Data Mining & Knowledge Management: Intelligent Query Processing

Application Category

• Security and Privacy: Data transparency and provenance
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Protecting patient privacy in clinical narratives is essential for enabling secondary use of healthcare data under regulations such as GDPR and HIPAA. While manual de-identification remains the gold standard, it is costly and slow, motivating the need for automated methods that combine privacy guarantees with high utility. Most automated text de-identification pipelines employed named entity recognition (NER) to identify protected entities for redaction. Although methods based on differential privacy (DP) provide formal privacy guarantees, more recently also large language models (LLMs) are increasingly used for text de-identification in the clinical domain. In this work, we present the first comparative study of DP, NER, and LLMs for Dutch clinical text de-identification. We investigate these methods separately as well as hybrid strategies that apply NER or LLM preprocessing prior to DP, and assess performance in terms of privacy leakage and extrinsic evaluation (entity and relation classification). We show that DP mechanisms alone degrade utility substantially, but combining them with linguistic preprocessing, especially LLM-based redaction, significantly improves the privacy-utility trade-off.

Problem

• differential privacy
• de-identification
• clinical notes
• privacy-utility trade-off
• Dutch

Innovation

• differential privacy
• large language models
• clinical text de-identification
• privacy-utility trade-off
• hybrid de-identification