From Answers to Arguments: Toward Trustworthy Clinical Diagnostic Reasoning with Toulmin-Guided Curriculum Goal-Conditioned Learning

TL;DR

This paper introduces a Toulmin-guided curriculum learning approach to improve the transparency and reliability of clinical diagnostic reasoning in large language models.

Contribution

It adapts the Toulmin model for clinical argumentation and proposes a novel curriculum training pipeline, CGCL, to enhance diagnostic reasoning in LLMs.

Findings

CGCL achieves diagnostic accuracy comparable to reinforcement learning methods.

The approach produces more transparent and structured clinical arguments.

Training with CGCL is more stable and efficient than resource-intensive methods.

Abstract

The integration of Large Language Models (LLMs) into clinical decision support is critically obstructed by their opaque and often unreliable reasoning. In the high-stakes domain of healthcare, correct answers alone are insufficient; clinical practice demands full transparency to ensure patient safety and enable professional accountability. A pervasive and dangerous weakness of current LLMs is their tendency to produce "correct answers through flawed reasoning." This issue is far more than a minor academic flaw; such process errors signal a fundamental lack of robust understanding, making the model prone to broader hallucinations and unpredictable failures when faced with real-world clinical complexity. In this paper, we establish a framework for trustworthy clinical argumentation by adapting the Toulmin model to the diagnostic process. We propose a novel training pipeline: Curriculum Goal-Conditioned Learning (CGCL), designed to progressively train LLM to generate diagnostic arguments that explicitly follow this Toulmin structure. CGCL's progressive three-stage curriculum systematically builds a solid clinical argument: (1) extracting facts and generating differential diagnoses; (2) justifying a core hypothesis while rebutting alternatives; and (3) synthesizing the analysis into a final, qualified conclusion. We validate CGCL using T-Eval, a quantitative framework measuring the integrity of the diagnosis reasoning. Experiments show that our method achieves diagnostic accuracy and reasoning quality comparable to resource-intensive Reinforcement Learning (RL) methods, while offering a more stable and efficient training pipeline.