It is argued that meaningful clinical impact will require the integration of multimodal learning, semantic knowledge representation, explainable reasoning, and human-in-the-loop decision processes to support safe, interpretable, and clinically deployable AI systems in pediatric healthcare environments.
Abstract
Artificial intelligence has emerged as a promising paradigm for advancing the screening, diagnosis support, and monitoring of autism spectrum disorder (ASD) through scalable and data-driven clinical augmentation. Recent advances in machine learning, multimodal sensing, and digital phenotyping have enabled AI systems to analyze behavioral, neurophysiological, speech, and clinical data to identify early markers of ASD. Despite encouraging experimental results, major barriers to clinical translation remain, including limited generalizability, fragmented datasets, insufficient evaluation rigor, lack of semantic interoperability, and unresolved ethical and regulatory concerns. This systematic review provides a comprehensive technical review of AI for ASD, covering data modalities, feature engineering, learning paradigms, evaluation protocols, deployment architectures, and knowledge representation frameworks. Particular emphasis is placed on system-level and translational considerations, including cloud–edge infrastructures, explainable clinical decision-support systems, privacy-aware deployment, and ontology-driven reasoning. Beyond summarizing existing work, this paper critically analyzes challenges related to reproducibility, dataset bias, interpretability, and clinical integration and derives design requirements for next-generation trustworthy ASD AI systems. We argue that meaningful clinical impact will require the integration of multimodal learning, semantic knowledge representation, explainable reasoning, and human-in-the-loop decision processes to support safe, interpretable, and clinically deployable AI systems in pediatric healthcare environments.
A comprehensive review of recent advancements in ASD research, with particular emphasis on neuroimaging, artificial intelligence (AI), and machine learning (ML)-based diagnostic approaches, highlights the growing potential of AI-driven tools for supporting early ASD diagnosis and emphasizes the need for standardized protocols, external validation, explainable AI, and clinically translatable frameworks.
Kuljeet Singh, Khushi Mogha, S. Moctar· Neurological Sciences· 0 citations
Based on the evaluated studies, transfer learning with diverse datasets and modalities has great promise for early ASD diagnosis, and a hybrid transfer learning-based framework is advised to assist clinicians and therapists in accurately diagnosing and assessing ASD severity.
R. Thillaikarasi, P. Kumaresan· International Conference on...· 0 citations
Artificial intelligence shows promise as a supportive tool for early screening, but current evidence supports its use as a complement to, rather than replacement for, clinical assessment.
Andrea Catalina Mahecha Ballesteros, Juanita Valeria García Bello, Eleaine Scarlet González Zuñiga et al.· Current Psychiatry Reports· 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
Artificial intelligence (AI) has rapidly emerged as a promising tool for improving autism spectrum disorder (ASD) screening, particularly in regions where access to trained specialists remains limited. Recent advances in machine learning have demonstrated encouraging diagnostic performance through the analysis of facial expressions, speech, eye gaze, and other behavioral markers, offering the potential to expand early diagnosis and reduce disparities in healthcare access. However, existing research has focused predominantly on algorithm development and diagnostic accuracy, while comparatively little attention has been paid to the broader ethical, cultural, and geopolitical implications of AI-assisted autism diagnosis. Drawing upon perspectives from philosophy of medicine and science and technology studies, this paper examines how AI redistributes power through three interconnected shifts: the transfer of diagnostic authority from clinicians to algorithms, the globalization of culturally specific definitions of "normal" behavior through Western-trained datasets, and the emergence of technological dependency on foreign-controlled AI infrastructure. Rather than rejecting AI-assisted diagnosis, this paper argues that equitable implementation requires culturally representative datasets, participatory model development, and greater local ownership of digital health technologies to ensure that AI promotes both diagnostic accessibility and global health equity.
Sophie Aoqing Qin· Theoretical and Natural Scie...· 0 citations
Introduction Early autism diagnosis remains challenging due to reliance on clinical observation and limited specialist availability. Addressing these barriers through automated diagnostic labeling and the integration of parental input may help mitigate the problem. Methods We trained a BioBERT machine learning model to label individual autism behavioral descriptions using the seven DSM-5 diagnostic criteria (A1-A3, B1-B4). This approach offers transparent clinical decision-making by providing detailed diagnostic information for individual behaviors and avoiding final case-level black-box decisions. We evaluated the model's performance on labeling lay (N = 35,971) and clinical (N = 145,603) behavior descriptions, as well as its transferability between the two. In addition, we compared the data sources by evaluating the diagnostic utility of lay and clinical examples across four dimensions, and of AI-generated summaries across two dimensions. Results We found that BioBERT can label both types of input, although it achieved higher precision (69%) on clinical descriptions and higher recall (83%) on lay descriptions. Sample size did not explain differences in performance. Transferring models from one data type to another results in a performance drop. Overall, training first on clinical data yielded the best-performing diagnostic models. When evaluating the examples from both data sources, the results show similar scores for the Utility, Specificity, Clinical Relevance, and Impact on Daily Life dimensions, and the cosine similarity analysis revealed substantial overlap (0.42) in vocabulary between the two. The utility of examples for A diagnostic behaviors was generally scored higher than that for B diagnostic behaviors. AI-generated summary scores showed a similar pattern between A and B examples but they were only moderately representative of these examples. Discussion These results demonstrate that lay behavioral descriptions can provide diagnostically valuable information comparable to clinical observations, although they are not readily summarized by AI. The integration of lay information into the diagnostic workflows could accelerate autism diagnosis without compromising clinical utility.
Gondy Leroy, Himanshu Nimbarte, Madhuri Sai Kandula et al.· Frontiers in Digital Health· 0 citations