Performance Assessment of ChatGPT-4.0 and ChatGLM Series in Traditional Chinese Medicine for Metabolic Associated Fatty Liver Disease: Comparative Study
This comparative study assesses the efficacy of ChatGPT-4.0 and ChatGLM models in diagnosing fatty liver disease, suggesting that TCM-specific training enhances decision support tools.
Key Points
ChatGLM4+Knowledge Base outperformed other models in syndrome differentiation and treatment principles, indicating its superior ability.
ChatGPT-4.0 exhibited confusion in 32 cases regarding TCM and Western medicine concepts, highlighting its limitations in specific contexts.
Analysis included 87 metabolic dysfunction-associated fatty liver disease cases treated with traditional Chinese medicine (TCM) for evaluation.
Real-time internet connectivity significantly improved LLMs' performance in TCM applications, emphasizing the need for pretraining with domain-specific knowledge.