Diabetes prediction method fusing knowledge expansion and convolutional neural network

A diabetes prediction method fusing knowledge extension and a convolutional neural network constructs a semantic enhanced convolutional neural network model taking a word embedding vector and a text embedding vector as dual-channel input, so that the model can pay attention to more important diabetes features and capture diabetes semantic information with finer granularity. Therefore, the diagnosis accuracy of diabetes is improved; the method solves the problems of strong dependence on large-scale labeling data in the diabetes mellitus field, lack of knowledge in the diabetes mellitus field and the like, a method for quoting external knowledge in the diabetes mellitus diagnosis process is designed, a data enhancement effect can be achieved on a data set, and data needed by deep learning model training is reduced; the problems that the deep learning model learning result generalization ability is not high due to the small scale of the medical data set, and a satisfactory diabetes prediction result cannot be ob