Closing Reasoning Gaps in Clinical Agents with Differential Reasoning Learning

2026-02-10

AI Summary

This work addresses the frequent logical gaps in clinical reasoning exhibited by AI agents due to a lack of clinical plausibility. To bridge this gap, we propose a Differential Reasoning Learning (DRL) framework that constructs directed acyclic graphs (DAGs) representing both agent-generated and reference clinical reasoning paths. By aligning corresponding nodes through a clinically weighted graph edit distance combined with an LLM-as-a-judge mechanism, the framework automatically generates natural language correction instructions. These corrections are then injected into the agent’s reasoning prompt via retrieval-augmented generation (RAG) to repair logical inconsistencies. Our approach is the first to integrate graph edit distance with LLM-based evaluation for clinical reasoning alignment, enabling the construction of a reusable differential reasoning knowledge base. Evaluated on medical question answering and follow-up prediction tasks, DRL significantly improves answer accuracy and reasoning trustworthiness under low token overhead, with effectiveness validated by clinical expert review.

Technology Category

• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Clinical decision support requires not only correct answers but also clinically valid reasoning. We propose Differential Reasoning Learning (DRL), a framework that improves clinical agents by learning from reasoning discrepancies. From reference reasoning rationales (e.g., physician-authored clinical rationale, clinical guidelines, or outputs from more capable models) and the agent's free-form chain-of-thought (CoT), DRL extracts reasoning graphs as directed acyclic graphs (DAGs) and performs a clinically weighted graph edit distance (GED)-based discrepancy analysis. An LLM-as-a-judge aligns semantically equivalent nodes and diagnoses discrepancies between graphs. These graph-level discrepancy diagnostics are converted into natural-language instructions and stored in a Differential Reasoning Knowledge Base (DR-KB). At inference, we retrieve top-$k$ instructions via Retrieval-Augmented Generation (RAG) to augment the agent prompt and patch likely logic gaps. Evaluation on open medical question answering (QA) benchmarks and a Return Visit Admissions (RVA) prediction task from internal clinical data demonstrates gains over baselines, improving both final-answer accuracy and reasoning fidelity. Ablation studies confirm gains from infusing reference reasoning rationales and the top-$k$ retrieval strategy. Clinicians'review of the output provides further assurance of the approach. Together, results suggest that DRL supports more reliable clinical decision-making in complex reasoning scenarios and offers a practical mechanism for deployment under limited token budgets.

Problem

• clinical reasoning
• reasoning gaps
• clinical decision support
• chain-of-thought
• reasoning fidelity

Innovation

• Differential Reasoning Learning
• Reasoning Graph
• Graph Edit Distance
• Retrieval-Augmented Generation
• Clinical Decision Support