Objective To evaluate the feasibility and screening accuracy of a brief scenario-based, video-delivered tool for early identification of autism spectrum disorder (ASD) in young children.Methods Data were analyzed from 211 children aged 12–42 months (ASD n=140, other developmental disorders OD n=35, and typically developing TD n=36) who completed a 5-minute scenario-based ASD early screening tool eliciting eight target behaviors: initiation/response to joint attention, response to name, eye contact, social referencing, imitation, social smiling, and pointing. Behaviors were scored using two criteria (0–16 symptom score; number of activities with partial or complete non-response). Diagnostic classification (ASD, OD, and TD) followed best-estimate diagnoses integrating Korean Version of the Autism Diagnostic Observation Schedule, Korean Version of the Autism Diagnostic Interview–Revised, Behavior Development Screening for Toddlers, Social Responsiveness Scale 2nd edition, Social Communication Questionnaire, and Korean Version of the Vineland Adaptive Behavior Scales–Second Edition. Group differences and screening performance were examined with analysis of variance, chi-square tests, and receiver operating characteristic analyses.Results Significant group differences emerged for response to name, eye contact, pointing, social referencing, social smiling, and initiation of joint attention, especially between ASD and non-ASD groups. Across scoring methods, children with ASD showed higher total scores and more non-responsive activities (all p<0.001). Area under the curve values were 0.703 for the total score and 0.676 and 0.700 for the two non-response indices, indicating good overall discrimination with relatively high sensitivity and modest specificity.Conclusion This scenario-based ASD early screening tool shows promising feasibility and accuracy as a brief, standardized video screener for toddlers and preschoolers. With its accessibility and scalability, it has potential for widespread use in community and home settings, warranting further refinement to improve specificity and implementation in real-world practice.
Bong et al. (2026) studied this question.