Abstract: Electronic medical records (EMR) play a critical role in modern healthcare. However, existing methods for quality control of electronic medical records, including traditional manual quality control and automated quality control systems, suffer from issues such as high workload, low efficiency, and limited coverage, making them inadequate for meeting the demands of efficient quality control in contemporary healthcare environments. To address these challenges, this paper proposes an EMR quality control system based on the DeepSeek large model, integrating prompt learning and knowledge-base-based retrieval-augmented generation (RAG) techniques to achieve automated quality control by leveraging multi-source medical knowledge. The experimental results demonstrate that this system has significantly optimized the omission rate, content deficiency rate, and quality control effectiveness of medical records. The missing filling rate of medical records has dropped from 9.42% to 3.55%, the connotative defect rate has decreased from 76.52% to 34.28%, and the medical record quality control rate has reached 100%. This study demonstrates the great potential of the DeepSeek-based automated quality control approach in enhancing both EMR quality and operational efficiency, providing strong support for the future intelligent development of EMR quality control.