Attention-guided CNN with Multi-stage Feature Selection for Explainable Autism Spectrum Disorder Detection

Read full text Summarise with VIRAT
Authors: Sahana Bisalapur, Geeta R. Bharamagoudar
Cited by: 0
References: 28

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Autism Spectrum Disorder (ASD) is a complex neurodevelopment disorder that is characterized by social communication and interactional issues, and behavioral disorders.It is important to note that ASD must be identified in their early and correct stages in order to give early intervention and secure improved developmental results.The traditional approaches to the diagnostics like clinical examination and questionnaires are biased and time-consuming and require the expert evaluation, but the general machine learning algorithms do not select features, do not provide interpretability, and do not extrapolate to non-homogenous data.To address these weaknesses, this paper proposes a deep learning model that incorporates Convolutional Neural Networks (CNN) and attention model.To increase stability of features and transparency of classification, the model uses systematic feature selection pipeline of Random Forest importance ranking, SelectKBest filtering, Recursive Feature Elimination (RFE), and SHAP-based interpretability analysis.Experiments were carried out on two publicly available datasets containing the UCI child dataset and the Kaggle toddler dataset using rigorous pre-processing, class balancing and cross-validation protocols.The described method achieves classification scores of 99.25 and 99.10 on the UCI and the Kaggle data set, respectively, which is higher than the existing machine learning models.The results demonstrate that the attentiondirected CNN mechanism is useful in the modeling of complicated behavioral patterns.

Fields of Study & Subjects

Autism Spectrum Disorder Research Artificial Intelligence in Healthcare Explainable Artificial Intelligence (XAI)

Publication details

Journal / source
International journal of intelligent engineering and systems
Volume
19
Issue
4
Pages
912–929
Publication date
23 March 2026
Language
EN