Construction of a machine learning-based model for stratified assessment of college students’ mental health and design of intervention pathways

This paper carries out more in-depth data mining and modeling as well as application work on the mental health of college students, and constructs a hierarchical assessment model of mental health of college students. The psychological data collected from students is normalized and balanced, and the construction of mental health characteristics is completed using personal, family, and school factors. Based on the Stacking framework, a two-layer multi-learners fusion algorithm is proposed, which integrates the SVM, RF, and XGBoost models to complete the setup of the base and meta-learners, and the different base learners are trained by using a 5-fold cross-validation method. 5,762 data of college students were collected and obtained through an online questionnaire to carry out the analysis of their mental health assessments. The number of students who were detected to have anxiety, human-computer relationship sensitivity and depression psychological state were 1303, 946 and 850 respectively, accounting for 22.62%, 16.42% and 14.76% of the total number of students. The number of students with severe anxiety symptoms can be up to 1303 and the number of students with more severe anxiety symptoms is 260. The number of students with depressive psychological states was 73.78% and 18.08%, respectively, of people with strong and more severe depression. There were 1274 students who displayed a sensitive or highly sensitive psychological state in interpersonal relationships. To address the mental health problems of students in higher education, targeted interventions can be carried out in terms of campus ecological interactions, individual precision interventions, and social practice services to enhance students’ mental health.