Enhanced Qwen-VL 7B Model via Instruction Finetuning on Chinese Medical Dataset

The integration of artificial intelligence (AI) into healthcare is revolutionizing medical information access and professional advice delivery. This study focuses on the development of a medical-specific question-answering model leveraging the Qwen-VL 7B model, an advanced Large Vision-Language Model (LVLM), to enhance the understanding and generation of medical texts. The Qwen-VL 7B model's capabilities in both language and visual comprehension make it an ideal candidate for medical question-answering systems, which require a deep grasp of medical knowledge and language. The research objectives include adapting the Qwen-VL 7B model to medical terminology, utilizing its visual understanding for medical imaging analysis, optimizing question-answering systems for medical scenarios, and evaluating the performance of these models. To achieve these goals, we employed methodologies such as dataset construction, model fine-tuning, and user studies. Results showed that the Qwen-VL-Mediacal model achieved a Rouge-1 score of 0.6147, indicating its potential for medical applications. However, challenges remain, such as understanding complex scenes and abstract concepts. Future research will aim to improve the model's adaptability and reasoning capabilities.