Jul 2026· SPU - Journal of Science, Technology and Management Research· 0 citations
TL;DR
A comparative analysis of recent approaches for early ASD prediction, focusing on methodologies, datasets, predictive performance, and key limitations indicates that deep learning-based approaches, particularly those using facial images and behavioral video analysis, generally achieve higher accuracy than traditional machine learning models.
Abstract
Early identification of Autism Spectrum Disorder (ASD) is crucial for enabling timely intervention and improving developmental outcomes. Conventional diagnostic methods rely heavily on behavioral observation and expert judgment, often leading to delayed diagnosis. In recent years, machine learning and deep learning techniques have been increasingly explored to support early ASD prediction using facial images, behavioral videos, neuro imaging data, and clinical datasets. This review presents a comparative analysis of recent approaches for early ASD prediction, focusing on methodologies, datasets, predictive performance, and key limitations. The findings indicate that deep learning-based approaches, particularly those using facial images and behavioral video analysis, generally achieve higher accuracy compared to traditional machine learning models. However, challenges such as limited dataset diversity, reduced cross-dataset generalization, lack of interpretability, and insufficient clinical validation remain prevalent across studies. In addition to the review, this work includes the implementation of an existing deep learning-based facial image classification method to validate reported findings. Overall, this critical review highlights current progress and emphasizes the need for robust, interpretable, and clinically validated approaches to support early ASD screening and complement traditional diagnostic practices.
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
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
Autism Spectrum Disorder (ASD) is a neurological and developmental condition characterized by challenges in social interaction, communication (both verbal and non-verbal), and repetitive behaviours. While genetics play a key role in its onset, early diagnosis remains essential for effective intervention. Machine learning (ML) offers a promising approach to streamline and accelerate ASD detection, making it faster and more cost-effective than traditional methods. This paper evaluates eight classification models to identify key ASD features and automate diagnosis. We compare their performance on large datasets to enhance predictive accuracy. ML has transformed healthcare by leveraging vast data volumes for analysis, with technological advances over the past decade improving diagnostic tools now standard in medical settings. ASD affects individuals variably, with symptoms typically appearing between 18 months and 3 years. Although genetic and environmental factors contribute, no single cause is confirmed. Traditional screenings rely heavily on clinician expertise, involving manual assessments and scoring, which can be subjective and time-consuming—even experts face uncertainties in predicting onset or severity. Parents seek rapid, reliable results. ML and deep learning (DL) address these gaps by analyzing complex patterns in data, enabling early prediction of ASD and its severity. This study implements diverse algorithms to support precise, automated screening, reducing diagnostic delays and improving outcomes.
Devireddy Mamatha, K. Maheswari· 2026 6th International Confe...· 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
Early diagnosis of autism spectrum disorder (ASD) plays a crucial role in facilitating prompt interventions and assessing post-therapy progress, both of which can greatly improve developmental trajectories. The emergence of artificial intelligence—particularly deep learning—has opened new possibilities for clinicians to detect ASD with improved precision and speed. This study investigates the utilization of deep learning models to recognize ASD among 13-year-old children based on eye movement data collected as participants observed static images and short video sequences. The dataset included visual and numerical variables, such as gaze position and pupil diameter, allowing for a multimodal analytical approach. For the numerical dataset, a multilayer perceptron (MLP) neural network produced the best outcomes, yielding an accuracy of 91.7% and a recall rate of 83.3% in ASD classification. Meanwhile, the Vision Transformer (ViT) model performed best for image-based analysis, reaching an accuracy of 78.2% and a recall rate of 88.6%. Overall, the findings highlight the potential of deep learning frameworks as objective, data-driven tools for ASD detection in both clinical and research contexts.Key words: Autism, Computer Vision, Deep Learning Model, Vision Transformer.
Muhamad Syukron, R. Faresta· Jurnal Ilmiah Kursor· 0 citations
Autism Spectrum Disorder (ASD) is a prevalent neurodevelopmental condition for which early screening is essential to support timely intervention. Conventional diagnostic procedures are often time-consuming and resource-intensive, highlighting the need for accessible and cost-effective screening approaches. This study investigates the use of deep learning–based facial image analysis as a supportive tool for ASD screening rather than as a diagnostic solution. A comparative evaluation of three state-of-the-art convolutional neural network architectures EfficientNetB0, MobileNetV2, and ResNet50 is conducted using a unified experimental framework based on transfer learning and fine-tuning. Experiments are performed on a publicly available Autistic Children Facial Image Dataset, with an additional verification step to prevent duplicate images in order to enhance generalization and reduce potential bias. All models are trained and evaluated under identical conditions to ensure a fair comparison. Experimental results indicate that EfficientNetB0 achieves the best overall performance, reaching a test accuracy of 88%, while maintaining a favorable balance between model complexity and generalization ability.