Autism Spectrum Disorder Detection Via Eye Gaze Analysis

Autism is a complex neuro - developmental condition characterized by challenges in social interaction and communication. Early diagnosis and intervention are crucial for effective support. This paper presents a novel approach for autism detection leveraging Convolutional Neural Networks (CNNs) to an...

Full description

Saved in:
Bibliographic Details
Published in2024 2nd International Conference on Networking and Communications (ICNWC) pp. 1 - 8
Main Authors Das, Sujay, Chotia, Akshat, Maheswari, K. M. Uma, A K, Sadiq
Format Conference Proceeding
LanguageEnglish
Published IEEE 02.04.2024
Subjects
Accuracy
Adaptation models
Autism
Autism Spectrum Disorder
CNN
Early Diagnosis
Evaluation Metrics
Feature extraction
Intervention
Loss
Measurement
MobileNetV2
Network topology
Neural networks
Neurodevelopmental Condition
Novel Approach
Relevant Features
Social Interaction
Training
Transfer learning
Online AccessGet full text

Cover

More Information
Summary:Autism is a complex neuro - developmental condition characterized by challenges in social interaction and communication. Early diagnosis and intervention are crucial for effective support. This paper presents a novel approach for autism detection leveraging Convolutional Neural Networks (CNNs) to analyze eye gaze patterns. The pro0posed method aims to provide an objective and efficient tool for the early screening of Autism. We begin by collecting eye gaze data from a diverse sample of individuals, including both Autism -diagnosed and non Autism individuals. The dataset is preprocessed to extract relevant features from the eye gaze sequences, capturing subtle but significant differences in gaze patterns between the two groups. A CNN architecture is designed and trained on this preprocessed dataset. The network is optimized to automatically learn discriminative features that distinguish between Autism and non Autism gaze patterns. Transfer learning techniques mobilenetv2 are employed to further enhance model performance. The trained CNN demonstrates promising results in accurately classifying individuals into Autism and neurotypical categories. Comprehensive evaluation metrics, including accuracy and loss, attest to the model's effectiveness.
DOI:10.1109/ICNWC60771.2024.10537260