对比大模型在临床命名实体识别中的表现，发现微调效果最佳。
Supervised Fine-Tuning or In-Context Learning? Evaluating LLMs for Clinical NER
- 用简单提示词进行上下文学习，比复杂指令更有效
- 监督微调使F1达87.1%，优于其他方法
- 大模型在二分类任务中表现更好，适合简化场景
我们在CADEC数据集上研究临床命名实体识别，比较三类方法：(i) BERT系列编码器（BERT Base、BioClinicalBERT、RoBERTa-large），(ii) GPT-4o结合少样本上下文学习（ICL）使用简单与复杂提示，(iii) GPT-4o监督微调（SFT）。所有模型在五个实体类型（ADR、Drug、Disease、Symptom、Finding）上评估。RoBERTa-large和BioClinicalBERT仅小幅优于BERT Base，显示该类模型的局限性。在大模型方案中，简单提示词的ICL优于复杂指令，而SFT整体性能最强（F1 ≈ 87.1%），但成本更高。我们发现大模型在简化任务（仅两类分类）中准确率更高。
原文摘要 · Abstract (English)
We study clinical Named Entity Recognition (NER) on the CADEC corpus and compare three families of approaches: (i) BERT-style encoders (BERT Base, BioClinicalBERT, RoBERTa-large), (ii) GPT-4o used with few-shot in-context learning (ICL) under simple vs.\ complex prompts, and (iii) GPT-4o with supervised fine-tuning (SFT). All models are evaluated on standard NER metrics over CADEC's five entity types (ADR, Drug, Disease, Symptom, Finding). RoBERTa-large and BioClinicalBERT offer limited improvements over BERT Base, showing the limit of these family of models. Among LLM settings, simple ICL outperforms a longer, instruction-heavy prompt, and SFT achieves the strongest overall performance (F1 $\approx$ 87.1%), albeit with higher cost. We find that the LLM achieve higher accuracy on simplified tasks, restricting classification to two labels.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。