TCM-Eval: An Expert-Level Dynamic and Extensible Benchmark for Traditional Chinese Medicine

TL;DR

This paper introduces TCM-Eval, a comprehensive benchmark for TCM, along with a new large-scale training corpus and a specialized LLM, ZhiMingTang, to advance AI applications in Traditional Chinese Medicine.

Contribution

It presents the first dynamic, extensible TCM benchmark, a novel data enrichment method SI-CoTE, and a TCM-specific LLM that surpasses human practitioner standards.

Findings

ZhiMingTang exceeds human practitioner passing thresholds

The SI-CoTE method effectively enriches reasoning chains

TCM-Eval facilitates standardized evaluation for TCM models

Abstract

Large Language Models (LLMs) have demonstrated remarkable capabilities in modern medicine, yet their application in Traditional Chinese Medicine (TCM) remains severely limited by the absence of standardized benchmarks and the scarcity of high-quality training data. To address these challenges, we introduce TCM-Eval, the first dynamic and extensible benchmark for TCM, meticulously curated from national medical licensing examinations and validated by TCM experts. Furthermore, we construct a large-scale training corpus and propose Self-Iterative Chain-of-Thought Enhancement (SI-CoTE) to autonomously enrich question-answer pairs with validated reasoning chains through rejection sampling, establishing a virtuous cycle of data and model co-evolution. Using this enriched training data, we develop ZhiMingTang (ZMT), a state-of-the-art LLM specifically designed for TCM, which significantly exceeds the passing threshold for human practitioners. To encourage future research and development, we release a public leaderboard, fostering community engagement and continuous improvement.