Large Language Models Versus Physicians in Traditional Chinese Medicine: A Real-World Clinical Case Evaluation

arXiv:2609.17544v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly being explored for clinical applications, yet their assessment for real-world traditional Chinese medicine (TCM) practice remains limited We constructed a clinical case library comprising 349 de-identified outpatient cases from 62 hospitals and evaluated 16 LLMs and a comparator cohort of 60 practicing TCM physicians using 60 representative cases selected from this library. Model outputs and physician reports were anonymized and scored by five senior TCM experts across nine diagnostic and therapeutic dimensions. Cutting-edge general-purpose LLMs achieved higher expert scores than the physician comparators, particularly for medical advice, treatment principles and selected diagnostic tasks. However, prescription-level analyses revealed discrepancies in herb selection, dosage, and treatment strategy, and qualitative safety review identified hallucinations and undesirable template-driven outputs. These findings highlight the potential of LLMs for TCM decision support while underscoring the need for physician oversight, safety constraints and prospective clinical evaluation.

LLMs show potential in TCM, but need oversight.

The study shows LLMs can outperform some physicians in certain TCM diagnostic tasks, indicating potential to improve healthcare efficiency. However, discrepancies in prescriptions and safety concerns like hallucinations mandate physician oversight, limiting autonomous deployment currently. The net effect suggests material clinical benefits balanced with significant risks.

This is a mixed, evidence-grounded clinical benchmark showing LLMs outperform physicians on some scored dimensions but fail on safety-critical prescription details. The explicit call for oversight and the documented hallucinations balance the promising decision-support potential, keeping the implication near neutral with slight positive lean.

The study presents a real-world evaluation where LLMs outperformed human physicians in certain aspects of TCM diagnosis and treatment planning. This demonstrates significant potential for AI in healthcare. However, identified safety concerns like hallucinations and prescription discrepancies temper the optimism, highlighting risks that need addressing.

SOURCE     arXiv cs.CY (Computers and Society) (tier 1)
   ↓
DOCUMENT   282fedeb-ed7c-4ae3-ac5b-ada73b893797
           https://arxiv.org/abs/2609.17544
   ↓
EVIDENCE   3 extracted excerpts
   ↓
MODEL RUN  4 runs, methodology 1.0.1
   ↓
SCORE      -8  (Mixed or uncertain)
   ↓
CONFIDENCE 76%