The University of Jordan :: A Convolutional Deep Neural Network Approach to Predict Autism Spectrum Disorder Based on Eye-Tracking Scan Paths

A Convolutional Deep Neural Network Approach to Predict Autism Spectrum Disorder Based on Eye-Tracking Scan Paths

Abstract:

This research investigates the application of a Convolutional Deep Neural Network (CDNN), specifically the T-CNN-ASD model, to analyze eye-tracking scan paths for early diagnosis of Autism Spectrum Disorder (ASD). The increasing global prevalence of ASD underscores the urgent need for non-invasive, efficient diagnostic tools. By leveraging advanced machine learning algorithms and eye-tracking data, this study presents a novel approach aimed at enhancing early intervention strategies and improving developmental outcomes for affected children.

Introduction:

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition characterized by significant social, communication, and behavioral challenges. The heterogeneity and complexity of ASD symptoms necessitate the development of accurate, non-invasive diagnostic tools. This research employs eye-tracking technology, a method that reflects the atypical visual attention patterns of individuals with ASD, to develop an innovative diagnostic model using deep learning techniques.

Methodology:

The study utilizes a comprehensive dataset comprising eye-tracking scan paths from children diagnosed with ASD and typically developing peers. The methodology encompasses data collection, image processing, and analysis through the T-CNN-ASD model. The research follows ethical standards and employs a cross-validation approach for model evaluation, ensuring the reliability and validity of the findings.

Results:

The T-CNN-ASD model demonstrated superior accuracy in classifying ASD and typically developing scan paths, outperforming traditional machine learning models. This section will detail the model's performance metrics, including accuracy, sensitivity, specificity, and F1 scores, providing a robust evaluation of its diagnostic capabilities.

Discussion and Expected Impact:

The findings underscore the potential of integrating eye-tracking technology with deep learning for ASD diagnosis. The T-CNN-ASD model's high accuracy offers promising directions for early ASD detection, which is crucial for timely intervention and support. The study discusses the implications for clinical practice, future research, and the development of personalized treatment plans.

Conclusion:

This research marks a significant step forward in the application of machine learning in the field of autism diagnosis. By providing an effective, non-invasive diagnostic tool, the T-CNN-ASD model holds the potential to revolutionize early detection and intervention strategies for ASD.

Project Team:

The study was conducted by a multidisciplinary team, bringing together expertise in artificial intelligence, and other related domains, highlighting the collaborative effort required to address complex health conditions like ASD.

References:

Alsaidi, M.; Obeid, N.; Al-Madi, N.; Hiary, H.; Aljarah, I. A Convolutional Deep Neural Network Approach to Predict Autism Spectrum Disorder Based on Eye-Tracking Scan Paths. Information 2024, 15, 133. https://doi.org/10.3390/info15030133

https://www.mdpi.com/2078-2489/15/3/133