In order to solve the problem of higher feature dimensions and sparse data that traditional feature extraction methods have in classification of disease symptoms

combining with the advantages of convolutional neural network（ CNN） and support vector machine（ SVM）

a disease symptom classification model of SVM based on CNN is proposed. The model firstly uses skip-gram to train the word vector according to the wiki Chinese corpus

which constitutes the two-dimensional feature matrix of disease symptom text

that is

input layer of CNN.Then

the text feature is extracted through convolution layer

and the local optimal feature of the text is obtained at the pooling layer using a 1-max pooling strategy. Finally

the local optimal feature composition fusion feature vector is used as the input of the SVM classifier to obtain the classification result. After comparison with the experimental results of the traditional feature extraction algorithm and the CNN algorithm

it is verified that the proposed model has significant improvement in three evaluation indicators of accuracy

recall rate and F1.