ORCID

0000-0002-9523-3346 (Jayawardena)

College

College of Sciences

Department

Computer Science

Graduate Level

Doctoral

Publication Date

2023

DOI

10.25883/29z1-ca84

Abstract

Processing speech with background noise requires appropriate parsing of the distorted auditory signal, fundamental language abilities as well as higher signal-to-noise ratio. Adolescents with ADHD have difficulty processing speech with background noise due to reduced inhibitory control and working memory capacity. In this study we utilize Audiovisual Speech-In-Noise performance and eye-tracking measures of young adults with ADHD compared to age-matched controls, and generate graphs for ADHD evaluation using the eye-tracking data. We form graphs utilizing the eight eye-tracking features (fixation count, average, total, and standard deviation of fixation duration, max and min saccade peak velocity, min, average, and standard deviation of saccade amplitude), and connection among trials in terms of subject, background noise, and sentence. We created multiple un-directed multi- graphs, each with 830 nodes which corresponds to a trial. Each trial is defined by a participant, background noise-level, and the sentence the participant was presented. For instance, k th node has information of {‘background noise level’: x, ‘sentence’: i, ‘subject’: j }. For each node, we create a feature matrix utilizing aforementioned eight eye gaze metrics. Links between pair of nodes mean that they belong to the same edge category. We introduced different types of edge categories: Same Background Noise Level, Same Subject, Same Sentence, Same Background Noise Level and Same Subject, Same Subject and Same Sentence, and Same Background Noise Level and Same Sentence. In our Graph Convolutional Network (GCN) model, we use node embedding and adjacency matrix representation as the input. The GCN layer is a multiplication of inputs, weights, and the normalized adjacency matrix. From the results we observed that only “Same Background Noise Level and Same Subject” edge category was able to give slightly better results in terms of AUC ROC and Precision. Additionally, we visualized what the model has learned by accessing the embeddings before the classification layer.

Keywords

Eye-tracking, Graph convolutional network, ADHD

Disciplines

Other Computer Sciences

Files

Download

Download Full Text (949 KB)

ADHD Prediction Through Analysis of Eye Movements With Graph Convolution Network


Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.