![[2503.12286] Integrating Chain-of-Thought and Retrieval Augmented Generation Enhances Rare Disease Diagnosis from Clinical Notes](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2503.12286] Integrating Chain-of-Thought and Retrieval Augmented Generation Enhances Rare Disease Diagnosis from Clinical Notes
Summary
This article presents a novel approach combining Chain-of-Thought (CoT) and Retrieval Augmented Generation (RAG) to improve rare disease diagnosis from unstructured clinical notes, demonstrating enhanced performance over traditional models.
Why It Matters
The integration of CoT and RAG addresses the significant challenge of utilizing unstructured clinical data for rare disease diagnosis, potentially improving patient outcomes by enabling more accurate gene prioritization and diagnosis in real-world clinical settings.
Key Takeaways
- Combining CoT and RAG enhances the analysis of clinical notes for rare disease diagnosis.
- Recent foundation models outperform earlier versions in candidate gene prioritization.
- RAG-driven CoT is effective for high-quality notes, while CoT-driven RAG excels with noisy data.
- The study utilized a diverse dataset, including Phenopacket-derived notes and clinical narratives.
- Improved accuracy in gene prioritization can lead to better clinical decision-making.
Computer Science > Computation and Language arXiv:2503.12286 (cs) [Submitted on 15 Mar 2025 (v1), last revised 18 Feb 2026 (this version, v2)] Title:Integrating Chain-of-Thought and Retrieval Augmented Generation Enhances Rare Disease Diagnosis from Clinical Notes Authors:Zhanliang Wang, Da Wu, Quan Nguyen, Kai Wang View a PDF of the paper titled Integrating Chain-of-Thought and Retrieval Augmented Generation Enhances Rare Disease Diagnosis from Clinical Notes, by Zhanliang Wang and 3 other authors View PDF Abstract:Background: Several studies show that large language models (LLMs) struggle with phenotype-driven gene prioritization for rare diseases. These studies typically use Human Phenotype Ontology (HPO) terms to prompt foundation models like GPT and LLaMA to predict candidate genes. However, in real-world settings, foundation models are not optimized for domain-specific tasks like clinical diagnosis, yet inputs are unstructured clinical notes rather than standardized terms. How LLMs can be instructed to predict candidate genes or disease diagnosis from unstructured clinical notes remains a major challenge. Methods: We introduce RAG-driven CoT and CoT-driven RAG, two methods that combine Chain-of-Thought (CoT) and Retrieval Augmented Generation (RAG) to analyze clinical notes. A five-question CoT protocol mimics expert reasoning, while RAG retrieves data from sources like HPO and OMIM (Online Mendelian Inheritance in Man). We evaluated these approaches on rare disease ...
