Scholarlink

<p>Adaptive digital therapies are needed with increasing rates of youth anxiety. However, existing systems lack personalization and dynamic learning. To solve this, the proposed research is to build a personalized ideological content recommendation model based on reinforcement learning to effectively reduce the anxiety of adolescents and young adults. The proposed system combines psychological, behavioral, emotional and contextual and engagement patterns for personalized therapeutic intervention. The proposed Efficient Henry Gas Solubility Optimized Deep Q-Networks (EHGSO-DQN) model uses optimization and Deep Reinforcement Learning (DRL) to enhance personalized therapeutic content recommendations for youth anxiety. The model adaptively recommends motivational, mindfulness-based, emotionally supportive, and Cognitive Behavioral Therapy (CBT) inspired therapeutic content based on user behavior and emotional interactions. The DQN method is used to learn adaptive therapeutic recommendation policies from emotional, behavioral, and contextual user data. The EHGSO optimization method improves convergence efficiency, recommendation accuracy, and personalization effectiveness for stable reinforcement learning performance. Longitudinal behavioral and mental health datasets collected from youth digital wellness platforms are utilized for training and evaluation. Data preprocessing using Min–Max Normalization and Principal Component Analysis (PCA) improves feature consistency and dimensionality reduction. Experimental results were obtained using a Python-based implementation environment. Experimental evaluation demonstrates that the reinforcement learning-driven recommendation model outperforms conventional collaborative filtering, static recommendation systems, and traditional deep learning approaches in personalization quality, therapeutic effectiveness, engagement retention, and anxiety reduction accuracy. Performance is assessed using97.65% accuracy, cumulative reward, and user satisfaction metrics. The findings demonstrate the potential of reinforcement learning-enabled digital therapeutics for scalable, adaptive, and emotionally intelligent youth mental healthcare systems.</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

Personalized ideological content recommendation via reinforcement learning as a digital therapeutic for anxiety reduction in youth

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月27日

查看全文
  • Yu Wang

摘要

Adaptive digital therapies are needed with increasing rates of youth anxiety. However, existing systems lack personalization and dynamic learning. To solve this, the proposed research is to build a personalized ideological content recommendation model based on reinforcement learning to effectively reduce the anxiety of adolescents and young adults. The proposed system combines psychological, behavioral, emotional and contextual and engagement patterns for personalized therapeutic intervention. The proposed Efficient Henry Gas Solubility Optimized Deep Q-Networks (EHGSO-DQN) model uses optimization and Deep Reinforcement Learning (DRL) to enhance personalized therapeutic content recommendations for youth anxiety. The model adaptively recommends motivational, mindfulness-based, emotionally supportive, and Cognitive Behavioral Therapy (CBT) inspired therapeutic content based on user behavior and emotional interactions. The DQN method is used to learn adaptive therapeutic recommendation policies from emotional, behavioral, and contextual user data. The EHGSO optimization method improves convergence efficiency, recommendation accuracy, and personalization effectiveness for stable reinforcement learning performance. Longitudinal behavioral and mental health datasets collected from youth digital wellness platforms are utilized for training and evaluation. Data preprocessing using Min–Max Normalization and Principal Component Analysis (PCA) improves feature consistency and dimensionality reduction. Experimental results were obtained using a Python-based implementation environment. Experimental evaluation demonstrates that the reinforcement learning-driven recommendation model outperforms conventional collaborative filtering, static recommendation systems, and traditional deep learning approaches in personalization quality, therapeutic effectiveness, engagement retention, and anxiety reduction accuracy. Performance is assessed using97.65% accuracy, cumulative reward, and user satisfaction metrics. The findings demonstrate the potential of reinforcement learning-enabled digital therapeutics for scalable, adaptive, and emotionally intelligent youth mental healthcare systems.