2026· ITEGAM- Journal of Engineering and Technology for Industrial Applications (ITEGAM-JETIA)· Vol 12, pp. 400-407· 0 citations
TL;DR
A toddler‑centered Case‑Based Reasoning (CBR) framework that emulates clinicians’ decision‑making by retrieving and adapting similar historical cases and delivers transparent, interpretable recommendations via comparable cases, supporting clinician trust is introduced.
Abstract
Early diagnosis of Autism Spectrum Disorder (ASD) in toddlers improves developmental outcomes through timely interventions. However, traditional methods rely on subjective judgment, lengthy assessments, and limited specialist availability. This study introduces a toddler‑centered Case‑Based Reasoning (CBR) framework that emulates clinicians’ decision‑making by retrieving and adapting similar historical cases. The system processes screening data from children aged 12–36 months, including demographics, risk factors, and three core behavioral domains: social interaction, communication, and repetitive behaviors. Cases are organized hierarchically by age and severity. The framework implements the full CBR cycle: structured representation, similarity‑based retrieval via clinically weighted Euclidean distance, DSM‑5–inspired rule‑guided adaptation, and dynamic maintenance through validation and pruning. On a public toddler ASD screening dataset, it achieves 84.2% accuracy, 79.3% sensitivity, 89.7% specificity, and an F1‑score of 0.82 using similarity‑weighted voting over top‑k cases, outperforming Random Forest (81.5% accuracy) and k‑NN (78.9% accuracy) baselines trained on the same features. It also delivers transparent, interpretable recommendations via comparable cases, supporting clinician trust. This clinically aligned CBR system provides an accessible AI tool for early toddler ASD diagnosis in healthcare settings.
Behavioral observation through examiner-child structured interactive play, particularly metrics related to eye contact, effectively differentiates toddlers with iASD.
Dan Ai, Binyue Hu, Qiuhong Wei et al.· BMC Psychiatry· 0 citations
Machine learning can support diagnosticians in this effort, as demonstrated here utilizing multiple rating scales, the TASI, and the TAP, but there is a risk for bias when using machine learning and as such, no algorithm should replace expert clinical judgment.
Aaron J. Kaat, Ashlynn Campagna, Hannah Feiner et al.· JCPP Advances· 0 citations
A Hybrid Intelligent Model designed to predict ASD in pediatric cases, leveraging adaptive neuro-fuzzy systems integrates artificial neural network capabilities with fuzzy logic, offering a comprehensive approach to ASD prediction.
Nneka MaryAnn Okafor, C. Ituma, R. Nweze· Communication in Physical Sc...· 0 citations
Selecting Evidence-Based Practices learning strategies for children with Autism Spectrum Disorder requires careful consideration of children’s diverse ability profiles. This study developed a classification model for Evidence-Based Practices learning strategies using the Random Forest algorithm based on a primary dataset consisting of 106 records. The features used include child age, gender, verbal ability, Autism Spectrum Disorder severity level, learning media, and language learning difficulties. The classification target consists of four classes: ABA, Visual Method, PECS, and Speech Therapy. Model evaluation was conducted using an 80:20 hold-out split, Stratified 5-Fold Cross Validation, Repeated Stratified Cross Validation, ablation test, and feature importance analysis. In the 80:20 hold-out scenario, Random Forest achieved an accuracy of 45.45% and an F1-Macro score of 0.4393. In Stratified 5-Fold Cross Validation, the model obtained an average accuracy of 38.61% ± 7.17% and an F1-Macro score of 0.3803 ± 0.0736. The repeated cross-validation results showed an average accuracy of 34.45% ± 9.46% and an F1-Macro score of 0.3332 ± 0.0967. These findings indicate that Random Forest is able to form an initial classification model; however, its performance remains low to moderate. The limitations of this study lie in the limited dataset size, the relatively small number of samples per class, and the absence of external validation; therefore, the model cannot yet be used as a final recommendation system. This study is positioned as a preliminary study to support the development of a Decision Support System for selecting Evidence-Based Practices learning strategies for children with Autism Spectrum Disorder.
S. Septiana, Dwi Krisbiantoro, Suliswaningsih Suliswaningsih· Jambura Journal of Electrica...· 0 citations
BACKGROUND
Early language delay is a common presenting concern among toddlers with autism spectrum disorder (ASD) and developmental language disorder (DLD), creating challenges for clinical assessment and referral decisions. Although ASD and DLD differ in core social-communicative features, restricted/repetitive behaviours, and developmental mechanisms, language-focused assessment may provide complementary information about early communicative profiles. However, culturally adapted tools for assessing early language and communication in Mandarin-speaking toddlers remain limited.
METHODS
A total of 213 children aged 24-36 months (ASD = 89, DLD = 60, TD = 64) were recruited from the Child Healthcare Department of a tertiary hospital between 2020 and 2022. Clinical diagnoses were established by multidisciplinary teams based on DSM-5 and CATALISE criteria. Caregivers completed the Diagnostic Receptive and Expressive Assessment of Mandarin-Infant & Toddler (DREAM-IT), which evaluates receptive and expressive language, cognitive play, and social communication. Receiver operating characteristic (ROC) curve analyses assessed discriminative validity across diagnostic groups, logistic regression examined the contribution of DREAM-IT domains to ASD-versus-DLD classification, and concurrent validity was evaluated against GDS, ABC, CARS, and ADOS-2.
RESULTS
DREAM-IT showed strong diagnostic accuracy in distinguishing children with ASD or DLD from TD children (AUC = 0.84-0.998), but its ASD-versus-DLD discrimination was domain-specific and more limited. Receptive language provided the strongest ASD-versus-DLD discrimination (AUC = 0.81; sensitivity = 69.1%; specificity = 84.3%; optimal cutoff = 86.5), whereas expressive language, cognitive play, and social communication showed poor-to-limited standalone performance (AUC = 0.51-0.57). Logistic regression indicated that receptive language (OR = 1.10, p < 0.05) and expressive language (OR = 0.77, p = 0.015) independent contributed to ASD-versus-DLD discrimination, with an overall classification accuracy of 79.2%. DREAM-IT classifications demonstrated moderate-to-substantial agreement with established developmental and autism-related instruments, including GDS, ABC, CARS, and ADOS-2 (Cohen's κ = 0.62-0.82). Developmental age comparisons revealed domain-specific lags, with ASD characterized by greater receptive and social delays and DLD showing the largest lag in expressive language.
CONCLUSIONS
The Mandarin DREAM-IT provides clinically useful information about early language and communication development in Mandarin-speaking toddlers with ASD and DLD and shows moderate-to-strong concurrent alignment with established developmental and autism-related measures. Its primary value lies in characterizing domain-specific developmental patterns and offering useful complementary information within a broader, multi-method diagnostic framework for Mandarin-speaking populations.
WHAT THIS PAPER ADDS
What is already known on this subject Differentiating autism spectrum disorder (ASD) from developmental language disorder (DLD) in toddlers is clinically important but challenging. Both conditions may present with early language delay, while ASD is characterised by core social-communication difficulties and restricted or repetitive behaviours. Accurate interpretation therefore requires a multi-method assessment integrating language evaluation, developmental history, direct observation, and autism-specific measures. However, culturally and linguistically appropriate tools for characterising early language and communication in Mandarin-speaking toddlers remain limited. What this study adds to the existing knowledge This study evaluated the diagnostic accuracy and concurrent validity of the Mandarin DREAM-IT in 213 toddlers with ASD, DLD, or typical development. DREAMIT showed good-to-excellent accuracy in distinguishing children with ASD or DLD from typically developing children, but ASD-versus-DLD discrimination was more limited and domain-specific. Receptive language provided the clearest discriminatory signal, whereas expressive language, cognitive play, and social communication showed substantial overlap. DREAM-IT developmental age scores showed moderate correlations with corresponding GDS measures, and its descriptive rule-based patterns showed substantial agreement with autism-specific instruments. What are the potential or actual clinical implications of this study? DREAM-IT may provide useful caregiver-reported information about receptive language, expressive language, cognitive play, and social communication in Mandarin-speaking toddlers. Its main clinical value lies in characterising domain-specific developmental patterns and providing complementary information within a broader, multi-method assessment framework. It should not be used as a standalone instrument to diagnose ASD or DLD or to differentiate between them. Results should be interpreted alongside developmental history, direct clinical observation, developmental testing, and autism-specific measures.
Jinzhu Zhao, Dandan Wu, Yutong Li et al.· International journal of lan...· 0 citations
In clinical practice, objectively distinguishing children with autism spectrum disorder (ASD) from those with typical development (TD) and other neurodevelopmental conditions with overlapping symptoms, such as global developmental delay (GDD), is critical for improving developmental outcomes. This study aimed to investigate the feasibility of eye-tracking-based machine learning (ML) models in distinguishing children with developmental concerns (ASD + GDD) from TD children, and in further differentiating ASD from GDD. A total of 168 toddlers (45 ASD, 56 GDD, and 67 TD; aged 18-48 months) viewed a socially dynamic "hide-and-seek" video. Both conventional area-of-interest (AOI)-based features and combined features derived from AOI combinations were used to train five ML classifiers. A two-stage classification framework was adopted, and SHapley Additive exPlanations (SHAP) analysis was used to interpret feature contributions. Combined features consistently outperformed isolated AOI features. For TD versus clinical classification, the random forest model achieved the highest accuracy of 80.36% (sensitivity = 84.16%, specificity = 74.63%). In differentiating ASD from GDD, the k-Nearest Neighbors model achieved the highest accuracy of 79.21% (sensitivity = 71.11%, specificity = 85.71%). SHAP analysis indicated that cross-regional attention features contributed substantially to model performance. These findings suggest that eye-tracking-based ML models may provide a promising approach for early screening of developmental conditions and ASD-GDD differentiation. Combined features reflecting cross-regional attention distribution demonstrated higher discriminative value than conventional AOI measures. Future studies should validate these findings in larger, multicenter, and more balanced samples with richer dynamic eye-tracking representations.
Gang Zhou, Xiaobin Zhang, Xingda Qu et al.· Research in Developmental Di...· 0 citations