![[2602.14160] Process-Supervised Multi-Agent Reinforcement Learning for Reliable Clinical Reasoning](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.14160] Process-Supervised Multi-Agent Reinforcement Learning for Reliable Clinical Reasoning
Summary
This paper presents a novel multi-agent reinforcement learning framework aimed at enhancing clinical reasoning by ensuring process-grounded decision-making alongside outcome accuracy.
Why It Matters
As clinical decision-making increasingly relies on AI, ensuring that these systems not only produce accurate outcomes but also adhere to established clinical reasoning processes is crucial. This research addresses a significant gap in current AI applications in healthcare, potentially improving patient outcomes and trust in AI-assisted clinical decisions.
Key Takeaways
- Introduces a process-supervised multi-agent reinforcement learning framework for clinical reasoning.
- Demonstrates improved outcome accuracy and process fidelity through dual reward systems.
- Highlights the importance of aligning AI decision-making with clinical standards.
Computer Science > Artificial Intelligence arXiv:2602.14160 (cs) [Submitted on 15 Feb 2026] Title:Process-Supervised Multi-Agent Reinforcement Learning for Reliable Clinical Reasoning Authors:Chaeeun Lee, T. Michael Yates, Pasquale Minervini, T. Ian Simpson View a PDF of the paper titled Process-Supervised Multi-Agent Reinforcement Learning for Reliable Clinical Reasoning, by Chaeeun Lee and 3 other authors View PDF HTML (experimental) Abstract:Clinical decision-making requires nuanced reasoning over heterogeneous evidence and traceable justifications. While recent LLM multi-agent systems (MAS) show promise, they largely optimise for outcome accuracy while overlooking process-grounded reasoning aligned with clinical standards. One critical real-world case of this is gene-disease validity curation, where experts must determine whether a gene is causally implicated in a disease by synthesising diverse biomedical evidence. We introduce an agent-as-tool reinforcement learning framework for this task with two objectives: (i) process-level supervision to ensure reasoning follows valid clinical pathways, and (ii) efficient coordination via a hierarchical multi-agent system. Our evaluation on the ClinGen dataset shows that with outcome-only rewards, MAS with a GRPO-trained Qwen3-4B supervisor agent substantially improves final outcome accuracy from 0.195 with a base model supervisor to 0.732, but results in poor process alignment (0.392 F1). Conversely, with process + outcome rewa...
