A depression recognition model based on multifeature fusion

In recent years, depression has become the primary problem of global mental health burden, and in order to identify it, a depression recognition model integrating BERT, BiLSTM and ConvNeXt is proposed. Firstly, the BERT model was used to generate feature vectors with rich semantics. Secondly, the BiLSTM model was used to obtain the context information of the text and the ConvNeXt model was used to obtain the local features of the text. Thirdly, in order to alleviate the loss of semantic information in the process of feature extraction, the context features and local features learned by the BiLSTM and ConvNeXt models were fused through residual connections. Finally, depression was identified based on the fused feature information. Experimental results show that compared with other deep learning models, the proposed model has improved accuracy, recall and F1 value, and can effectively extract depression features in the text and improve the accuracy of depression recognition.