A multisource data-driven mental health assessment model for vocational college students

Abstract

The frequent occurrence of crimes and suicide incidents among college students due to psychological abnormalities has become a hot topic in social media. Therefore, this article proposes a big data based psychological health assessment model for vocational college counselors and students. The input data for the model includes student campus card behavior data, social media text emotional features, and academic performance change trajectories. The training labels of the model are derived from the clinical diagnosis records of the school's psychological counseling center and the joint evaluation results of standardized psychological assessment scales. The core of the model is a backpropagation neural network (BPNN) model optimized based on particle swarm optimization (PSO) algorithm. In terms of risk assessment mechanism, the model quantifies students' mental health status as a risk index between 0-1, and automatically triggers three-level warnings (blue attention, yellow warning, red intervention) based on preset thresholds (such as>0.7). Finally, the warning information and key attention list are pushed to the counselor through a visual dashboard. The results indicate that the model can effectively achieve accurate assessment of college students' mental health, with high practicality and reliability.