SHIELD: 一个多样化的临床笔记数据集和蒸馏小语言模型，用于企业级去标识化
SHIELD: A Diverse Clinical Note Dataset and Distilled Small Language Models for Enterprise-Scale De-identification
AI总结：
针对现有临床去标识化基准数据集的不足，构建了包含1381份笔记、10229个PHI标注的多样化数据集SHIELD，并通过教师-学生蒸馏框架将大语言模型能力迁移至可本地部署的小模型，在标准工作站上达到0.89精确率和0.88召回率。
摘要：
临床文本的去标识化是电子健康记录二次利用的前提。现有的公共基准（如i2b2 2006和2014语料库）已超过十年，缺乏现代临床叙述的语义和人口多样性。大语言模型（LLMs）达到了最先进的零样本提取性能，但其在企业规模的使用受到计算成本和医院数据治理的限制，后者禁止将受保护健康信息（PHI）发送到云API。我们引入了SHIELD（用于学习和去标识化的合成人工标注标识符替换条目），这是一个多样化的临床笔记数据集，包含1381份笔记，涵盖9个类别的10229个金标准PHI跨度，通过集合覆盖多样性抽样（跨越人口统计和文档类型层）和人工参与裁决构建。我们评估了四个LLM（两个专有，两个开放权重）以在SHIELD上建立性能上限，然后展示了教师-学生蒸馏框架将这些能力转移到可本地部署的小语言模型中。我们最好的蒸馏模型在标准工作站硬件上运行时达到了微平均跨度级精确率0.89和召回率0.88。它在每个类别的召回率上落后于其云教师（宏平均0.90 vs. 0.81），但考虑到其较低的成本和本地部署能力，仍然具有竞争力。跨数据集评估表明，多样性训练的模型在通用结构化PHI类别上泛化良好，而机构特定的实体在双向转移中仍然困难，这表明将广泛覆盖模型与专门模型配对用于高容量、半结构化笔记类型。我们公开发布了SHIELD数据集和蒸馏的DeBERTa v3模型，以提供一种完全部署在机构防火墙内的准确、经济高效的去标识化流水线。
英文摘要：
De-identification of clinical text is a prerequisite for the secondary use of electronic health records. Existing public benchmarks such as the i2b2 2006 and 2014 corpora are over a decade old and lack the semantic and demographic diversity of modern clinical narratives. Large Language Models (LLMs) reach state-of-the-art zero-shot extraction, but their use at enterprise scale is limited by computational cost and by hospital data governance that restricts sending Protected Health Information (PHI) to cloud APIs. We introduce SHIELD (Synthetic Human-annotated Identifier-replaced Entries for Learning and De-identification), a diverse clinical note dataset of 1,381 notes with 10,229 gold-standard PHI spans across 9 categories, built with set-cover diversity sampling across demographic and document-type strata and human-in-the-loop adjudication. We evaluate four LLMs (two proprietary, two open-weight) to establish a performance ceiling on SHIELD, then show that a teacher-student distillation framework transfers these capabilities into locally deployable Small Language Models. Our best distilled model reaches micro-averaged span-level precision of 0.89 and recall of 0.88 while running on standard workstation hardware. It trails its cloud teacher on per-category recall (0.90 vs. 0.81 macro-averaged) but remains competitive given its lower cost and on-premise deployability. Cross-dataset evaluation shows that diversity-trained models generalize well on universal structured PHI categories, while institution-specific entities remain hard to transfer in both directions, which suggests pairing broad-coverage models with specialized models for high-volume, semi-structured note types. We publicly release the SHIELD dataset and the distilled DeBERTa v3 model to provide an accurate, cost-effective de-identification pipeline deployable entirely behind institutional firewalls.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
