Validation-Guided Transformer–Ensemble Framework for Early Autism Spectrum Disorder Screening Using Q-CHAT-10 Screening Data

Arnab Dutta; Arnab Barman; Rajib Das; Ritu Kotal; Avranil Sarkar

doi:10.36948/ijfmr.2026.v08i05.89369

Validation-Guided Transformer–Ensemble Framework for Early Autism Spectrum Disorder Screening Using Q-CHAT-10 Screening Data

Author(s)	Mr. Arnab Dutta, Mr. Arnab Barman, Mr. Rajib Das, Ms. Ritu Kotal, Mr. Avranil Sarkar
Country	India
Abstract	Autism spectrum disorder (ASD) is a neurodevelopmental condition for which early identification enables timely behavioral intervention, yet conventional diagnosis relies on lengthy clinical assessment that is scarce in low-resource settings. Using the public Q-CHAT-10 toddler screening dataset (1,054 records), and motivated by a recent TPOT-based automated machine learning (AutoML) study on a comparable locally collected dataset from Pakistan that reached an accuracy of 78% with an F1-score of 86% for the autistic class, this paper investigates whether a neural classifier trained with validation-based model selection can improve screening performance. The model with the best validation score (epoch 6) was evaluated on a held-out test set with a fixed decision threshold of 0.5. It achieved an accuracy of 97.48%, precision of 96.49%, recall of 100%, F1-score of 98.21%, specificity of 91.84%, and ROC-AUC of 1.0000. No autistic child in the test set was missed, while a small number of non-autistic children were flagged for follow-up. We discuss these results against the AutoML baseline and classical machine learning models, and we critically examine threats to validity, in particular the fact that the class label in this dataset is a deterministic function of the questionnaire items, the modest sample size, and the demographic composition of the data. The results suggest that questionnaire-based neural screening is a promising, low-cost triage tool, but external validation is required before clinical use.
Keywords	Autism Spectrum Disorder, Early Detection, Q-CHAT-10, Deep Learning, Automated Machine Learning, Screening, Healthcare
Field	Engineering
Published In	Volume 8, Issue 5, September-October 2026
Published On	2026-10-10
DOI	https://doi.org/10.36948/ijfmr.2026.v08i05.89369

View / Download PDF File

About IJFMR Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us		+91-9687-828-838	editor@ijfmr.com

International Journal For Multidisciplinary Research

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Bi-monthly Scholarly International Journal

Validation-Guided Transformer–Ensemble Framework for Early Autism Spectrum Disorder Screening Using Q-CHAT-10 Screening Data

Share this