Scholarlink

Attention Deficit Hyperactivity Disorder (ADHD) has two symptoms: inattention and hyperactivity. Inattention is the inability of a person to concentrate on a specific task fully. Hyperactivity/Impulsivity, on the other hand, is characterized by excessive activity and difficulty in physically and emotionally restraining oneself. Artificial Intelligence (AI) techniques have been implemented to assist these individuals, and among these methods the Reinforcement Learning (RL). Therefore, this paper aims to use RL to address the case of children with ADHD. We begin by presenting some cognitive deficits of ADHD patients. Then, we mention studies that have utilized it in the context of ADHD and we discuss the principles of RL. Next, we collect a sample of 106 children and adolescents with or without ADHD. We propose a Model-Free RL methodology using Q-Learning, SARSA, and a hybrid version of the two. Among the findings, we noted that 5.2% of inattentive children with ADHD are dysgraphic, 8.2% of those in this category focus on a task for less than 10 min, 9.3% of hyperactive children walk very quickly, and 4.4% of those in this category eat biscuits. The results also showed that for 300 episodes, the Q-Learning and Combined algorithms converge the most with an alpha of 0.9 and 0.5, respectively, and with a gamma of 0.1 and 0.5, respectively.

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

Cure-Free: A Free-Model Reinforcement Learning Approach for the ADHD Children

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Zineb Namasse,
  • Zineb Hidila,
  • Mohamed Tabaa,
  • Mounia Elhaddadi,
  • Samar Mouchawrab

摘要

Attention Deficit Hyperactivity Disorder (ADHD) has two symptoms: inattention and hyperactivity. Inattention is the inability of a person to concentrate on a specific task fully. Hyperactivity/Impulsivity, on the other hand, is characterized by excessive activity and difficulty in physically and emotionally restraining oneself. Artificial Intelligence (AI) techniques have been implemented to assist these individuals, and among these methods the Reinforcement Learning (RL). Therefore, this paper aims to use RL to address the case of children with ADHD. We begin by presenting some cognitive deficits of ADHD patients. Then, we mention studies that have utilized it in the context of ADHD and we discuss the principles of RL. Next, we collect a sample of 106 children and adolescents with or without ADHD. We propose a Model-Free RL methodology using Q-Learning, SARSA, and a hybrid version of the two. Among the findings, we noted that 5.2% of inattentive children with ADHD are dysgraphic, 8.2% of those in this category focus on a task for less than 10 min, 9.3% of hyperactive children walk very quickly, and 4.4% of those in this category eat biscuits. The results also showed that for 300 episodes, the Q-Learning and Combined algorithms converge the most with an alpha of 0.9 and 0.5, respectively, and with a gamma of 0.1 and 0.5, respectively.