HSNMF enables accurate and effective analysis for the college students' psychological health education data and student life data

Introduction College students face different levels of anxiety, depression, and other psychological problems due to various factors such as academic stress, excess workload, and family responsibilities. The state of mind plays a crucial role in shaping individuals' daily behaviors and academic performance. To comprehensively analyze the psychological health status of college students and research domains related to psychological health education, it is urgently needed to develop effective tools and models. Methods In this study, we proposed a novel framework called hypergraph-induced semi-orthogonal nonnegative matrix factorization (HSNMF). By using this framework, we can effectively evaluate the college students' psychological health levels. Results We implemented the proposed algorithm on two real datasets, and the results showed that the proposed algorithm outperformed other competing methods. The identified research domains provided insights into psychological health education. We also implemented a depression-level classification task on the student life dataset. The results showed that the low-dimensional latent variables learned from HSNMF contained rich semantic information, further improving the performance of traditional machine learning models. Clustering and regression analyses performed on the student life dataset showed that the depression status of students was significantly correlated with their performance in class and social life, as indicated by variables such as “Number of friends ( p -value = 0.000598),” “Gender ( p -value = 0.000034),” and “Taking notes in class ( p -value = 0.03).” Discussion The significance of student psychological health study is discussed.