Integration of Knowledge Graph and CNN-GRU in College Students' Mental Health Education and Psychological Crisis Intervention

Abstract

To improve the predictive ability of depression behavior among college students and improve the mental health warning system for college students, this study proposes a support vector machine prediction model with a depression knowledge graph and a depression prediction model with a convolutional neural network gated loop unit, according to the Weibo platform and language, gender, behavior, and emotional characteristics, to predict depression behavior among college students. The model constructed by the research institute outperformed the comparison model in all indicators, with an average improvement of 9.95%, 12.2%, 14.55%, and 11.55% in accuracy, recall, F1 score, and precision. In addition, the area value under the operating characteristic curve of the subjects in the final prediction model tended to 0.872, which was an average improvement of 0.059. The accuracy, recall, F1 score, and precision of the proposed prediction model had been improved by an average of 0.216, 0.140, 0.169, and 0.081. Thus, the model constructed by the research institute exhibits superior predictive ability for depressive behavior, providing theoretical reference for optimizing the mental health warning system for college students.