Research on school social work methodology based on predictive control and architectural innovations

Abstract

Abstract This study focuses on innovations in school social work methods and explores applying predictive control-based models to effectively support and improve students’ mental health and behavioral development. The XGBoost algorithm and support vector machine were employed to predict students’ mental health. The study was influenced by physiological, self, family, school, social, and network factors, and a questionnaire was developed based on the SCL-90 Symptom Self-Rating Scale. A validity rate of 94.32% was achieved by collecting 432 valid questionnaires. The results indicated a significant positive correlation between the influencing factors and students’ mental health, with physiological and family factors being the most influential. An AUC value of 93.6% and an accuracy rate of 93.9% were achieved by the XGBoost-SVM model when predicting students’ mental health. The prediction model can update students’ mental health status in real-time and provide an effective predictive control tool for school social work.