Entity Decomposition with Filtering: A Zero-Shot Clinical Named Entity Recognition Framework

2024-07-05

arXiv.org

AI Summary

This work addresses three key challenges in zero-shot clinical named entity recognition (NER): fine-grained entity omission, class imbalance, and low recall for rare/long-tail entities. To this end, we propose the Entity Decomposition and Filtering (EDF) framework—the first of its kind—decoupling open-domain NER into two stages: subtype-aware retrieval and collaborative result filtering. EDF leverages open-source, NER-specialized large language models and integrates task decomposition with type-aware mechanisms. Extensive experiments across multiple clinical benchmarks demonstrate that EDF consistently outperforms all baseline methods across all evaluation metrics, model configurations, and entity types. Notably, it achieves substantial improvements in the recognition accuracy of rare and long-tail clinical entities, significantly enhancing zero-shot generalization capability.

Technology Category

• Natural Language Processing: Information Extraction
• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion
• Machine Learning: Mixture of Experts (MoE)

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Clinical named entity recognition (NER) aims to retrieve important entities within clinical narratives. Recent works have demonstrated that large language models (LLMs) can achieve strong performance in this task. While previous works focus on proprietary LLMs, we investigate how open NER LLMs, trained specifically for entity recognition, perform in clinical NER. Our initial experiment reveals significant contrast in performance for some clinical entities and how a simple exploitment on entity types can alleviate this issue. In this paper, we introduce a novel framework, entity decomposition with filtering, or EDF. Our key idea is to decompose the entity recognition task into several retrievals of entity sub-types and then filter them. Our experimental results demonstrate the efficacies of our framework and the improvements across all metrics, models, datasets, and entity types. Our analysis also reveals substantial improvement in recognizing previously missed entities using entity decomposition. We further provide a comprehensive evaluation of our framework and an in-depth error analysis to pave future works.

Problem

• Evaluating open NER LLMs in clinical entity recognition
• Introducing entity decomposition with filtering framework
• Improving recognition of missed clinical entities

Innovation

• Open NER LLMs for clinical recognition
• Entity decomposition improves retrieval accuracy
• Filtering enhances sub-type entity recognition