Developing a Machine Learning Algorithm to Assess Attention Levels in ADHD Students in a Virtual Learning Setting using Audio and Video Processing
Srivi Balaji1, Meghana Gopannagari2, Svanik Sharma3, Preethi Rajgopal4

1Srivi Balaji, Meridian World School, Round Rock, United States of America.
2Meghana Gopannagari, Thomas Jefferson School of Science and Technology, Alexandria, United States of America.
3Svanik Sharma, Vandegrift High School, Leander, United States of America.
4Preethi Rajgopal, Kelley School of Business, Indiana University – Bloomington, Bloomington, United States of America.

Manuscript received on May 21, 2021. | Revised Manuscript received on May 28, 2021. | Manuscript published on May 30, 2021. | PP: 285-295 | Volume-10 Issue-1, May 2021. | Retrieval Number: 100.1/ijrte.A59650510121 | DOI: 10.35940/ijrte.A5965.0510121
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Over the past few years, numerous technological advancements have modernized and eased access to educational materials, improving overall learning experiences for students with ADHD despite the transition to remote learning. However, the majority of these improvements address comprehension and practice outside of the classroom without recognizing the need for engagement during a lesson. Students are more likely to retain higher amounts of information outside of class, if they have a strong understanding of the lesson during class. A back-end model combined with an engaging front-end user interface can enhance the standard of education for students with ADHD and help them achieve the same level of understanding they would have during an in-person lesson. This project aimed to address the remote learning experiences of students with ADHD by creating a model using machine learning to analyze audio and video clips of a live online lesson, detect distractions in the student’s environment, and use this data in tandem with an interactive user interface to engage students and enhance their remote learning experience. The two means of data collection employed in this model were audio and video analysis. This data was fed into separate convolutional neural networks with reinforcement learning architecture to identify distractions. A genetic algorithm was used to weigh the outputs of both neural networks and produce coefficients determining the weight of each factor. This was then used to determine the distraction level of the student. This model can be implemented in a virtual lesson between an instructor and a student with ADHD, to constantly monitor the attention level of the student. Findings of this research suggested that this model could help an instructor acknowledge and manage symptoms of ADHD – which may lead to distractions, such as impulsivity, hyperactivity and boredom – by modifying their curriculum to further engage the student. This research has the potential to fill the notable gap between technology and education, using technology to improve online educational quality for students with ADHD.
Keywords: ADHD, genetic algorithm, machine learning, neural networks, virtual education.