Large Language Models in Rare Disease Diagnosis and Clinical Decision-Making

The field of clinical natural language processing is moving towards the integration of large language models (LLMs) in the analysis of rare diseases and clinical decision-making. This direction is driven by the potential of LLMs to uncover insights and patterns from textual data, enabling the formulation of accurate and timely diagnoses. The use of LLMs in clinical settings is also being explored, with a focus on evaluating their performance in tasks such as dense information extraction from clinical case reports, clinical reasoning, and decision-making in emergency rooms. Noteworthy papers in this area include: CaseReportBench, which introduces a novel benchmark dataset for dense information extraction from clinical case reports and demonstrates the effectiveness of LLMs in extracting clinically relevant details. CDR-Agent, which presents a novel LLM-based system for autonomously identifying and applying clinical decision rules based on unstructured clinical notes, achieving significant accuracy gains relative to standalone LLM baselines. Second Opinion Matters, which proposes a consensus mechanism that enables improved clinical decision-making through the ensemble of specialized medical expert agents, demonstrating consistent accuracy gains across multiple medical evaluation benchmarks.

Sources

Decoding Rarity: Large Language Models in the Diagnosis of Rare Diseases

CaseReportBench: An LLM Benchmark Dataset for Dense Information Extraction in Clinical Case Reports

ER-REASON: A Benchmark Dataset for LLM-Based Clinical Reasoning in the Emergency Room

Case-Based Reasoning Enhances the Predictive Power of LLMs in Drug-Drug Interaction

CDR-Agent: Intelligent Selection and Execution of Clinical Decision Rules Using Large Language Model Agents

Second Opinion Matters: Towards Adaptive Clinical AI via the Consensus of Expert Model Ensemble

Built with on top of