Scholarlink

<p>The mental well-being of university students is increasingly recognized as an important concern because it can influence students’ academic performance, social functioning, and overall quality of life. Current intervention practices are based on assessments that use self-report measures and manual counseling platforms, which can be inconsistent, time-consuming, and less personalized. Established work lacks immediate, real-time data-based recommendations and doesn’t offer transparent explanations for recommendations that diminish the act of intervention. This study proposes Psychological Strategy Generation with Explainable Artificial Intelligence (Psi-GenXAI) as a new, innovative framework that integrates neural networks with explainable AI approaches. The proposed framework uses demographic variables, behavioural trends, psychological assessments, and digital engagement to generate predictions for levels of risk for stress, anxiety, and depression. To increase the transparency of prediction rationale, explainable AI methods SHAP and LIME were incorporated to provide reasoned interpretations of outcome predictions, thereby increasing trust in model recommendations. The Psi-GenXAI would enable personalized psychological interventions for students based on a risk continuum (education workshops, peer counselling, and therapy referrals). The experimental study demonstrates that Psi-GenXAI predicted student mental health risks while providing practically useful and understandable explanations, enabling informed clinician decision-making and early intervention. Psi-GenXAI achieves an accuracy of 91.2% and an F1 score of 0.98, demonstrating success in predicting students’ emotional and behavioural outcomes. Explainability analysis indicated that the contribution of features was consistent by 87.5% across validation sets, thus allowing counsellors to learn why each intervention was recommended. According to the analysis of multidimensional effects, the intervention recommendation module of Psi-GenXAI can positively change student support outcomes, with estimated indicators of a 28% increase in emotional well-being, a 24% decrease in behavior risk indicators, a 15% increase in engagement, and a 14% increase in academic performance under model-guided support.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A method for generating psychological intervention strategies for college students based on neural networks and explainable AI

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月12日

查看全文
  • Huayan Li

摘要

The mental well-being of university students is increasingly recognized as an important concern because it can influence students’ academic performance, social functioning, and overall quality of life. Current intervention practices are based on assessments that use self-report measures and manual counseling platforms, which can be inconsistent, time-consuming, and less personalized. Established work lacks immediate, real-time data-based recommendations and doesn’t offer transparent explanations for recommendations that diminish the act of intervention. This study proposes Psychological Strategy Generation with Explainable Artificial Intelligence (Psi-GenXAI) as a new, innovative framework that integrates neural networks with explainable AI approaches. The proposed framework uses demographic variables, behavioural trends, psychological assessments, and digital engagement to generate predictions for levels of risk for stress, anxiety, and depression. To increase the transparency of prediction rationale, explainable AI methods SHAP and LIME were incorporated to provide reasoned interpretations of outcome predictions, thereby increasing trust in model recommendations. The Psi-GenXAI would enable personalized psychological interventions for students based on a risk continuum (education workshops, peer counselling, and therapy referrals). The experimental study demonstrates that Psi-GenXAI predicted student mental health risks while providing practically useful and understandable explanations, enabling informed clinician decision-making and early intervention. Psi-GenXAI achieves an accuracy of 91.2% and an F1 score of 0.98, demonstrating success in predicting students’ emotional and behavioural outcomes. Explainability analysis indicated that the contribution of features was consistent by 87.5% across validation sets, thus allowing counsellors to learn why each intervention was recommended. According to the analysis of multidimensional effects, the intervention recommendation module of Psi-GenXAI can positively change student support outcomes, with estimated indicators of a 28% increase in emotional well-being, a 24% decrease in behavior risk indicators, a 15% increase in engagement, and a 14% increase in academic performance under model-guided support.