Large Language Models Perform Diagnostic Reasoning

Cheng-Kuang Wu, Wei-Lin Chen, Hsin-Hsi Chen
7/18/2023
cs.CL

No official code found. Best implementation: nlplab-best-team/diagnostic-reasoning (4 stars, last commit 2023-07-28).

Abstract

We explore the extension of chain-of-thought (CoT) prompting to medical reasoning for the task of automatic diagnosis. Motivated by doctors' underlying reasoning process, we present Diagnostic-Reasoning CoT (DR-CoT). Empirical results demonstrate that by simply prompting large language models trained only on general text corpus with two DR-CoT exemplars, the diagnostic accuracy improves by 15% comparing to standard prompting. Moreover, the gap reaches a pronounced 18% in out-domain settings. Our findings suggest expert-knowledge reasoning in large language models can be elicited through proper promptings.

View on arXivView PDF

Code Implementations(1)

nlplab-best-team/diagnostic-reasoningCommunity100%

Explore the ability of large language models (LLMs) to perform history taking through diagnostic reasoning, and how to enhance such ability.

40Python, Jupyter NotebookFeb 3, 20233 years ago

Similar Papers

Diagnostic Reasoning Prompts Reveal the Potential for Large Language Model Interpretability in Medicine

88% match2023

An automatically discovered chain-of-thought prompt generalizes to novel models and datasets

85% match2023

Large Language Models are Clinical Reasoners: Reasoning-Aware Diagnosis Framework with Prompt-Generated Rationales

85% match2023

Guiding Clinical Reasoning with Large Language Models via Knowledge Seeds

85% match1 repo2024

MedCausalX: Adaptive Causal Reasoning with Self-Reflection for Trustworthy Medical Vision-Language Models

85% match2026

ThoughtSource: A central hub for large language model reasoning data

85% match1 repo1.0k2023

Cite this paper

@article{wu2023large,
  title  = {Large Language Models Perform Diagnostic Reasoning},
  author = {Cheng-Kuang Wu and Wei-Lin Chen and Hsin-Hsi Chen},
  year   = {2023},
  eprint = {2307.08922},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2307.08922v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2307.08922.svg)](https://codeofpaper.com/paper/2307.08922)

Discussion