Early Detection of Autism Spectrum Disorder Using AI and Machine Learning Models: A Systematic Review for Effective Intervention

Autism is a brain developmental condition that may impact the everyday life of affected individuals. Toddlers suffering from ASD are labeled by difficulties in social interaction, connecting, lack of eye contact, ignoring danger, preferring to play alone, sensitivity to loud noises, and repetitive behavior compared to normally developed (ND) children. This systematic review examines the application of artificial intelligence (AI) and machine learning (ML) models for the early detection of Autism Spectrum Disorder (ASD), emphasizing their potential for enabling timely and effective interventions. The significance of this research lies in addressing the critical need for early ASD diagnosis, which can substantially improve outcomes for affected individuals. The novelty of this study is its comprehensive analysis of cutting-edge AI and ML techniques specifically applied to ASD detection, providing insights into their efficacy and limitations. The methodology involved a systematic search of peer-reviewed literature published between 2010 and 2022 across major databases, including PubMed, IEEE Xplore, and Scopus. Studies were selected based on predefined inclusion criteria, focusing on AI and ML models used for early ASD detection. Key findings reveal that AI and ML models, particularly deep learning algorithms and ensemble methods, demonstrate promising results in early ASD detection, with some models achieving accuracy rates exceeding 95%. The review identified that behavioral data, neuroimaging, and genetic information are the most commonly used input features for these models. However, challenges persist in terms of model interpretability, dataset bias, and generalizability across diverse populations. This systematic review highlights the potential of AI and ML models as valuable tools for early ASD detection, potentially revolutionizing diagnostic processes and enabling earlier interventions. Future research should focus on developing more robust, interpretable models and validating them across larger, more diverse datasets to enhance their clinical applicability and reliability in real-world settings.