With the popularization of Electronic Health Records (EHR), the emergency system has stored a large number of historical dispatch records, which can provide valuable insights for the optimization of current pre-hospital care. However, the inconvenient interaction manner of cur-rent information retrieval systems hinders researchers from exploring these historical records. To address this issue, we propose a novel framework that leverages the language understanding and code generation ability of Large Language Models (LLMs) to build an information retrieval system with Visualization-oriented Natural-language-based Inter-faces (V-NLI). To incorporate both domain-specific and task-related prior knowledge, we generate the instruction datasets based on the ability of closed-source LLMs in a multi-stage manner and conduct supervised fine-tuning on open-source LLMs. We also devised various mechanisms for augmenting the capabilities of open-source LLMs in query interpretation and code generation. To validate the effectiveness and generalizability of our framework, we conducted experiments on a public dataset NLV. More significantly, we performed more detailed experiments on a dataset including over 1 mil-lion pre-hospital emergency historical records in ten years. The performance of our method surpasses all baseline methods and achieves comparable results even with some SOTA closed-source models.

Traditional Chinese Medicine (TCM) presents unique challenges for Large Language Models (LLMs) due to its complex diagnostic reasoning. We introduce ZhongJingGPT, a specialized LLM for TCM that integrates vertical domain fine-tuning strategies with cognitive psychology insights. Our approach incorporates multi-TCM scenario and knowledge instruction construction strategies, enhanced by symptom sequence-based beam search and a Medical Finite State Machine (MedicalFSM) module. Using only Low-Rank Adaptation (LoRA) fine-tuning, ZhongJingGPT achieves state-of-the-art accuracy, surpassing Generative Pre-trained Transformer 4 (GPT-4) in key TCM-specific accuracy and fluency metrics. Comprehensive evaluations, including out-of-distribution assessments, renowned TCM practitioners’ case studies, and multi-turn role-playing scenarios, verify its superior performance on Chinese Massive Multitask Language Understanding (CMMLU) and TCM humanities datasets. A multi-dimensional evaluation standard, assessed by professional practitioners, further validates its effectiveness. This research demonstrates the potential of specialized LLMs in TCM and offers insights for Artificial Intelligence (AI) development in complex professional domains, bridging ancient wisdom with modern artificial intelligence technologies.