To help doctors better diagnose diseases,overcome the lack of systematic and effective means in the process of clinical diagnosis and treatment and make the best of the medical knowledge in electronic medical records,a diagnosis model is proposed based on visual analysis and data mining for electronic medical record.Firstly,electronic medical records of fever of unknown origin are preprocessed into structured data by extracting patients'symptoms.Secondly,the structured data is organized and visualized based on specific requirements.Finally,a diagnosis model is trained to discover the relationship between symptoms and causes,helping doctors find the potential medical knowledge in medical records and assisting doctors to diagnose.A visual analysis and auxiliary diagnosis framework for clinical diagnosis and treatment is designed based on the above analysis.Experiments show that the system could help doctors analyze the knowledge of electronic medical records of unknown cause,which could help doctors diagnose diseases in a shorter period of time.