Scholarlink

Mental health challenges among adults aged 18–40 are shaped by academic, occupational, and social stressors. This study surveyed 200 participants (130 males, 70 females) to analyze demographic, behavioral, and psychological indicators influencing well-being. Four machine learning algorithms-Random Forest, XGBoost, Logistic Regression, and Support Vector Regression-were applied to classify disorders such as anxiety, depression, OCD, PTSD, and sleep/eating disorders. Model performance was evaluated using accuracy, precision, recall, F1-score, and Top-3 accuracy. XGBoost achieved the highest performance (70% accuracy, 95% Top-3 accuracy), outperforming baselines. Behavioral indicators such as emotional exhaustion, panic attacks, mood swings, and suicidal ideation were the most predictive features. Limitations include gender imbalance and reliance on self-reports. Findings highlight AI’s supportive-not diagnostic-role in early detection and intervention [1] work also provides a theorical foundation for applying AI Techniques in relapse prediction, emphasizing its potential role in sustained mental health management.

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

AI-Driven Mental Health Prediction and Therapy for Early Adulthood: Relapse Prediction After Initial Treatment Response

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Chittin Arora,
  • Pooja Gupta,
  • Trilok Suthar

摘要

Mental health challenges among adults aged 18–40 are shaped by academic, occupational, and social stressors. This study surveyed 200 participants (130 males, 70 females) to analyze demographic, behavioral, and psychological indicators influencing well-being. Four machine learning algorithms-Random Forest, XGBoost, Logistic Regression, and Support Vector Regression-were applied to classify disorders such as anxiety, depression, OCD, PTSD, and sleep/eating disorders. Model performance was evaluated using accuracy, precision, recall, F1-score, and Top-3 accuracy. XGBoost achieved the highest performance (70% accuracy, 95% Top-3 accuracy), outperforming baselines. Behavioral indicators such as emotional exhaustion, panic attacks, mood swings, and suicidal ideation were the most predictive features. Limitations include gender imbalance and reliance on self-reports. Findings highlight AI’s supportive-not diagnostic-role in early detection and intervention [1] work also provides a theorical foundation for applying AI Techniques in relapse prediction, emphasizing its potential role in sustained mental health management.