A Human-Centric Pipeline for Aligning Large Language Models with Chinese Medical Ethics

TL;DR

This paper introduces MedES, a Chinese medical ethics benchmark and a guardian-in-the-loop framework to align large language models with complex medical ethical standards, improving their performance in ethical decision-making tasks.

Contribution

It presents a novel, scenario-centric benchmark and a guardian-in-the-loop alignment pipeline tailored for Chinese medical ethics, enhancing LLM ethical compliance.

Findings

Aligned model outperforms larger baselines on ethical tasks

Automated evaluator achieves over 97% accuracy in domain-specific assessments

Framework is adaptable to other legal and cultural environments

Abstract

Recent advances in large language models have enabled their application to a range of healthcare tasks. However, aligning LLMs with the nuanced demands of medical ethics, especially under complex real world scenarios, remains underexplored. In this work, we present MedES, a dynamic, scenario-centric benchmark specifically constructed from 260 authoritative Chinese medical, ethical, and legal sources to reflect the challenges in clinical decision-making. To facilitate model alignment, we introduce a guardian-in-the-loop framework that leverages a dedicated automated evaluator (trained on expert-labeled data and achieving over 97% accuracy within our domain) to generate targeted prompts and provide structured ethical feedback. Using this pipeline, we align a 7B-parameter LLM through supervised fine-tuning and domain-specific preference optimization. Experimental results, conducted entirely within the Chinese medical ethics context, demonstrate that our aligned model outperforms notably larger baselines on core ethical tasks, with observed improvements in both quality and composite evaluation metrics. Our work offers a practical and adaptable framework for aligning LLMs with medical ethics in the Chinese healthcare domain, and suggests that similar alignment pipelines may be instantiated in other legal and cultural environments through modular replacement of the underlying normative corpus.