Exploring Gaze Pattern Differences Between ASD and TD Children Using Internal Cluster Validity Indices

2024-09-18

AI Summary

Conventional autism spectrum disorder (ASD) screening relies on subjective behavioral assessments and manually defined areas of interest (AOIs), limiting objectivity and scalability. Method: This study proposes, for the first time, unsupervised clustering validity indices—including silhouette coefficient and Calinski-Harabasz index—as biologically interpretable biomarkers for differentiating eye-movement patterns between children with ASD and typically developing (TD) peers. Using three public eye-tracking datasets, we applied seven clustering algorithms (e.g., K-means, DBSCAN) to fixation trajectories and extracted 63-dimensional cluster-structure quality features, which were fed into XGBoost and SVM classifiers. Contribution/Results: Our approach eliminates reliance on handcrafted AOIs while preserving interpretability through principled, data-driven cluster evaluation. Empirical results demonstrate statistically significant associations between clustering validity metrics and neurodevelopmental status. In cross-dataset evaluation, the model achieves an AUC of 81%, establishing a novel, objective, and data-driven paradigm for early ASD screening.

Technology Category

• Machine Learning: Clustering
• Humans and AI: Brain-Sensing and Analysis
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Normalization, clustering, classification, and summarization of Web text
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Autism Spectrum Disorder (ASD) affects children's social and communication abilities, with eye-tracking widely used to identify atypical gaze patterns. While unsupervised clustering can automate the creation of areas of interest for gaze feature extraction, the use of internal cluster validity indices, like Silhouette Coefficient, to distinguish gaze pattern differences between ASD and typically developing (TD) children remains underexplored. We explore whether internal cluster validity indices can distinguish ASD from TD children. Specifically, we apply seven clustering algorithms to gaze points and extract 63 internal cluster validity indices to reveal correlations with ASD diagnosis. Using these indices, we train predictive models for ASD diagnosis. Experiments on three datasets demonstrate high predictive accuracy (81% AUC), validating the effectiveness of these indices.

Problem

• Distinguish gaze patterns ASD TD children
• Use internal cluster validity indices
• Train predictive models ASD diagnosis

Innovation

• Internal cluster validity indices
• Unsupervised clustering algorithms
• Predictive models for ASD diagnosis