With the continuous development of China’s medical modernization technology, medical devices have been widely used in the prevention, diagnosis and treatment of diseases, and health care. The direct or indirect contact of medical devices with the human body is closely related to diagnosis, treatment and rehabilitation. As a result, the quality and safety problems of medical devices in the development of high-end, high-tech products have also increased. Establishing an effective quality management system for medical device production can effectively control the level of production quality management in medical devices and guarantee the quality of products. Therefore, the management of the quality of medical devices is a major task. The use of digital construction to manage scientific and technological archives can make the management of scientific and technological archives more standardized, automatic and professional. This is also a development trend of hospital science and technology archives management. In this paper, we use pattern recognition technology and multimedia vision technology to study the construction of medical technology management system. The deep learning is integrated to promote the overall performance of the proposed methodology. The simulation results have reflected the robustness.