Scholarlink

<p>Mental health is a crucial aspect of overall well-being, yet it is often overlooked due to stigma and limited accessibility to care. This study investigates the ability of artificial intelligence (AI) to predict common psychological conditions, depression, anxiety, and stress, using validated psychometric data. We analyzed responses from the Depression Anxiety Stress Scales-42 (DASS-42) questionnaire, combined with demographic information, drawn from a large publicly available dataset of 39,&#xa0;775 anonymized participants. Five machine learning models were evaluated: decision tree, random forest, k-nearest neighbor, naive Bayes, and support vector machine (SVM). Data preprocessing included handling missing values, demographic standardization, and validity checks. Model performance was assessed using stratified train-test splits and five-fold cross-validation. The SVM model achieved the highest accuracy (99.3% for depression, 98.9% for anxiety, 98.8% for stress). These findings highlight the potential of AI-based approaches for early mental health screening, although further clinical validation is necessary to ensure their real-world applicability.</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

Artificial intelligence for predicting depression anxiety and stress using psychometric data

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月24日

查看全文
  • Tamer ShamsEldin,
  • Sarah Gaber,
  • Shuja Ansari,
  • Rania Elgohary,
  • Mahmoud A. Shawky,
  • Magdy Elbahnasawy,
  • Mohammed Abdrabou

摘要

Mental health is a crucial aspect of overall well-being, yet it is often overlooked due to stigma and limited accessibility to care. This study investigates the ability of artificial intelligence (AI) to predict common psychological conditions, depression, anxiety, and stress, using validated psychometric data. We analyzed responses from the Depression Anxiety Stress Scales-42 (DASS-42) questionnaire, combined with demographic information, drawn from a large publicly available dataset of 39, 775 anonymized participants. Five machine learning models were evaluated: decision tree, random forest, k-nearest neighbor, naive Bayes, and support vector machine (SVM). Data preprocessing included handling missing values, demographic standardization, and validity checks. Model performance was assessed using stratified train-test splits and five-fold cross-validation. The SVM model achieved the highest accuracy (99.3% for depression, 98.9% for anxiety, 98.8% for stress). These findings highlight the potential of AI-based approaches for early mental health screening, although further clinical validation is necessary to ensure their real-world applicability.