Multi-view Neural Network Integrating Knowledge for Patient Self-diagnosis

The electronic medical records contain a wealth of information, and are used in many medical tasks such as medical diagnosis. Most of the research are to assist doctors in diagnosis, and few studies are based on patient self-diagnosis. Our work is completely from the patient's point of view, through the patient's symptoms and discomfort body parts to determine the patient's possible disease. We have designed a multi-view neural network to fully capture the characteristics of multiple aspects of the patient, then perform feature fusion, and finally achieve the purpose of predicting disease only through the patient's symptoms and body parts. At the same time, we create a medical knowledge graph based on the patient's electronic medical record data. The facts in knowledge graph can effectively screen out the candidate disease of the patient, reduce the range of disease selection, and effectively improve the accuracy of the prediction. The experimental results also confirmed the effectiveness of the modified method.