Scholarlink

<p>Autism spectrum disorder (ASD) is a neurodevelopment disorder characterized by difficulties in social skills, repetitive behaviors, speech, and nonverbal communication. While there are various diagnostic methods used for ASD assesses, this work focuses on the Autism Quotient Trait 10 (AQ-10) which is a non-clinical diagnostic method that assesses autism in children, adolescents, and adults using 10 questions. However, the AQ-10 in its current form can only detects the presence of ASD without providing additional information. To improve the AQ-10’s detection capacity, we propose a new model based on AQ-10 that assigns weights to each question, as opposed to the initial model that assumes all questions have the same influence on the results. Our model, named as Mahi’s model, improves the number of classes from 10 to an interval between − 20 and 60, allowing for precise tracking of autism variations within the same class. We employ an Artificial Neural Network to estimate weight matrices and train our model on an open database containing 509 cases of individuals of different ages and genders of child. The results show a strong correlation between the output of our model and the real data. The results obtained demonstrate the potential of our new model to enhance early identification and support for individuals with autism spectrum disorder.</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

New improved model for Autism Detection and Prediction based on AQ-10 test by using Artificial Neural Network

  • 期刊论文

  • 发表日期：

    2024年12月31日

查看全文
  • Mahi Sarra,
  • Imedjdouben Fayçal,
  • Ridouh Abdelhakim

摘要

Autism spectrum disorder (ASD) is a neurodevelopment disorder characterized by difficulties in social skills, repetitive behaviors, speech, and nonverbal communication. While there are various diagnostic methods used for ASD assesses, this work focuses on the Autism Quotient Trait 10 (AQ-10) which is a non-clinical diagnostic method that assesses autism in children, adolescents, and adults using 10 questions. However, the AQ-10 in its current form can only detects the presence of ASD without providing additional information. To improve the AQ-10’s detection capacity, we propose a new model based on AQ-10 that assigns weights to each question, as opposed to the initial model that assumes all questions have the same influence on the results. Our model, named as Mahi’s model, improves the number of classes from 10 to an interval between − 20 and 60, allowing for precise tracking of autism variations within the same class. We employ an Artificial Neural Network to estimate weight matrices and train our model on an open database containing 509 cases of individuals of different ages and genders of child. The results show a strong correlation between the output of our model and the real data. The results obtained demonstrate the potential of our new model to enhance early identification and support for individuals with autism spectrum disorder.