Evidence is provided that mobile video-based ASD diagnosis can achieve comparable performance to models trained on clinical instrument data, and contributes to the development of broader adaptable autism detection tools, bypassing the dependence on traditional clinical instrument data.
Abstract
Background: Digital behavioral phenotyping of autism spectrum disorder (ASD) offers a promising approach for developing more scalable diagnostic frameworks across diverse global contexts. Machine learning (ML) models show promise for ASD diagnosis using behavioral videos, but critical questions remain regarding whether models trained on data from one country work in another, and how the background of the raters affects the accuracy. Our work addresses these questions by testing whether ML models can accurately diagnose ASD across different populations and rater groups. Methods: This work evaluates the performance of a supervised ML framework for binary classification of ASD versus non-ASD [speech, language and communication disorders (SLC) + neurotypical (NT)] in a cohort of 227 children in Bangladesh. We first assessed the cross-domain model transferability of a clinical-instrument-trained logistic regression model (LR-9) on behavioral ratings that were based on videos of Bangladeshi children interacting with caregivers and toys at two major child development centers in Dhaka, Bangladesh. We then trained five diverse classifiers (Logistic Regression, Random Forest, XGBoost, SVM, and RuleFit) on the full annotated Bangladeshi dataset. Using SHAP-based consensus elbow feature selection, we identified a compact set of features that maintained the performance. Finally, we developed ensemble models to improve predictive stability. Results: The LR-9 model, originally trained on U.S. clinical instrument data, was evaluated on video-based behavioral ratings from 214 Bangladeshi children. When tested on Bangladeshi clinician ratings, the LR-9 model achieved a sensitivity of 86.1% (95% CI: [0.78–0.93]) and AUC of 0.79 (95% CI: [0.73–0.86]). The distinction across rater groups was between trained raters (clinicians and students) and crowd workers, who showed lower sensitivity 28.5% (95% CI: [0.21, 0.39]). When tested on the aggregated ratings from all groups, the model achieved an AUC of 0.78 (95% CI: [0.72–0.84]). Inter-rater reliability followed the same pattern: individual agreement was fair (Krippendorff’s α = 0.26), but the multi-rater consensus was reliable (ICC(1,k) = 0.84), with Bangladeshi clinicians showing the highest agreement (α = 0.34) and crowd workers the lowest (α = 0.20). We then trained new models directly on the Bangladeshi ratings. All model types achieved similar AUC values (0.86–0.89), with overlapping confidence intervals. Using just 8–11 key behaviors kept the similar performance while cutting the features by 66–75%. Combining ensembles gave similar results (e.g., Bayesian averaging: AUC 0.88 [0.78, 0.95]) but with more stable predictions. Conclusion: This study provides evidence that mobile video-based ASD diagnosis can achieve comparable performance (AUC: 0.89 [0.76, 0.96]) to models trained on clinical instrument data. This work contributes to the development of broader adaptable autism detection tools, bypassing the dependence on traditional clinical instrument data.
Autism Spectrum Disorder (ASD) is a heterogeneous neurodevelopmental condition defined by differences in social communication and restricted, repetitive behaviours. As diagnostic criteria have broadened, ASD is now recognised across a wider range of individuals, raising key questions about its structure: does ASD have discrete sub-types, or is it better conceptualised as a continuous, possibly multidimensional, condition? We aim to explore whether a multidimensional continuum model more accurately captures the variability within ASD. We analysed a large SPARK phenotypic dataset of medical history and diagnostic surveys (background history, SCQ, RBS-R; n=36,710 individuals). We apply and compare two traditional statistical approaches, Factor Analysis and Gaussian Mixture Models, with a modern machine learning technique, the Variational Autoencoder (VAE). VAEs reconstructed unseen test data with ~4-fold better accuracy than Factor Analysis, and ~8-fold better accuracy than Gaussian Mixture Models. We identified four stable latent factors across 100 independently trained VAEs. These four dimensions provide an individual behavioural profile that can be visualized using radar-plots, offering a compact way to compare profiles at the person level. Through further analysis, we found evidence for 3 overlapping clusters or subtypes of ASD identified within the 4D latent space. This work aims to inform new ways of modelling ASD using a VAE that will be able to discern between a continuum or a clustered output and that go beyond binary diagnosis, instead reflecting the complex range of trait profiles, with implications for personalised diagnosis and intervention.
H. Quigley, B. Gardiner, L. McDaid et al.· medRxiv· 0 citations
Autism spectrum disorder (ASD) is a heterogeneous neurodevelopmental condition characterized by diverse behavioral, cognitive, sensory, and communication profiles, making early diagnosis and personalized intervention challenging. Recent advances in machine learning (ML) and deep learning (DL) have enabled the development of computational tools for ASD screening, classification, severity assessment, and intervention monitoring. This review synthesizes findings from 50 recent studies that applied ML and DL techniques to ASD-related datasets, including electroencephalography (EEG), eye-tracking, behavioral video, microbiome, voice acoustic, demographic, and multimodal data. The review addresses three key questions: (i) which data modalities and computational approaches are most frequently used, (ii) how diagnostic performance is evaluated across different study designs, and (iii) what methodological challenges limit clinical translation. The literature is organized according to data modality, algorithmic approach, and clinical readiness. Approaches examined include conventional ML methods, convolutional neural networks, graph neural networks, hybrid deep learning architectures, federated learning, explainable artificial intelligence, topological data analysis, and multimodal fusion. The findings suggest that multimodal and graph-based approaches provide a more comprehensive representation of ASD phenotypes than single-modality methods. Explainability and privacy-preserving learning have also emerged as important considerations for clinical deployment. However, many reported high-performance models are based on small sample sizes, repeated use of the ABIDE dataset, class imbalance, single-site validation, or limited external testing, raising concerns regarding generalizability. Beyond diagnostic accuracy, this review evaluates model interpretability, calibration, scalability, validation rigor, and clinical applicability. Overall, the analysis highlights the need for standardized benchmarks, externally validated multimodal datasets, clinically relevant evaluation metrics, and decision-support systems that complement rather than replace expert clinical assessment in ASD diagnosis and management.
S. K, Lakshmi Annapurna Y· Journal of Visualized Experi...· 0 citations
Autism Spectrum Disorder (ASD) classification using machine learning has shown promising results on behavioral screening datasets, however, such performance may be influenced by embedded questionnaire scoring and threshold-based patterns. This study presents an explainable multi-model ASD classification Framework that consist of questionnaire based screening, augmentation analysis, rule based interpretation, explainable artificial intelligence, and rs-fMRI based neuroimaging classification. The proposed framework is organized into three progressive models. ASD classification on child behavioral dataset focuses on child behavioral screening data and evaluated Logistic Regression, XGBoost, Linear Support Vector Machine, and Bernoulli Naive Bayes with XAI and rule-based interpretation. Multi-cohort ASD classification framework combined child, adolescent, and and adult AQ_10 based datasets and assessed the same classifiers under real, SMOTE, ADASYN, CTGAN, and TVAE training conditions. ABIDE-1 multi-modal ASD classification extended the analysis to ABIDE-1 rs-fMRI functional connectivity and phenotypic features using ElasticNet Logistic Regression, Linear SVM, Ridge Classifier, and SGD LogLoss classifier. The questionnaire based models achieved strong performance, with ASD classification on child behavioral dataset framework obtaining a best accuracy of 97% using Linear SVM and Multi-cohort ASD classification framework reaching up to 100% accuracy in selected real and traditional oversampling settings for logistic Regression and linear SVM. SMOTE and ADASYN produced more stable performance than CTGAN and TVAE. ABIDE-1 multi-modal ASD classification framework achieved a best accuracy of 70% using ElasticNet Logistic Regression, reflecting the grater complexity of neuroimaging based ASD classification.
Children and adolescents with ASD exhibited lower empathy capabilities than control subjects, which may be attributed to dysfunctions in the salience and social brain networks.
Yong-Lu Wang, Zhang-Liang Ma, Zhiyi Wang et al.· Frontiers in Psychiatry· 0 citations
The need to develop large, well‐balanced datasets, the application of explainable AI techniques, standardization and regulatory guidelines for facilitating the clinical translation of ASD detection systems are suggested.
Anupama N, Chandrashekar M. Patil· International Journal of Dev...· 0 citations
It is demonstrated that lay behavioral descriptions can provide diagnostically valuable information comparable to clinical observations, although they are not readily summarized by AI.
Gondy Leroy, Himanshu Nimbarte, Madhuri Sai Kandula et al.· Frontiers in Digital Health· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
