A Medical Consultation System Based on Federated Learning Framework

There exists a widely concerned social phenomenon in recent Chinese society. Chinese doctors and patients have formed a medical conflict relationship. To solve this tense relationship, this paper introduced a medical consultation system based on federated learning and retrieval augmented generation LLM. This system applies federated learning to ensure data privacy while transferring among different institutions. The multimodal knowledge graph integrates various medical data such as text, images, and signals, enhancing the model's diagnostic and therapeutic capabilities. Retrieval augmented generation (RAG) combines the generative capabilities of large models with real-time retrieval of external knowledge bases, improving the accuracy of answers. In addition, the system also uses an expert allocation mechanism to improve the quality of questions and answers. Experimental results show that the system performs well on the BLEU evaluation metric, providing high-quality medical consultation services.