Empowering Early Autism Detection in Toddlers Using Deep Neural Networks

Early detection of Autism Spectrum Disorder (ASD) in toddlers is crucial for preventing its severity and mitigating long-term effects. This study utilizes Deep Neural Networks (DNN) to extract complex patterns and meaningful features from extensive datasets to enable early autism detection. This research utilizes a diverse dataset encompassing facial images of children and behavioral characteristics sourced from online platforms. Various Deep Neural Network (DNN) architectures, including Multi-Task Cascaded Convolutional Networks and EfficientNet, are evaluated to analyze the complex data patterns. The application of these advanced DNN models shows the potential to revolutionize the diagnostic process, enabling earlier interventions. The proposed approach not only supports early intervention but also helps to overcome the challenges faced by healthcare systems and families. By considering multiple aspects of early childhood behavior, these models can effectively differentiate the children with ASD and those without, paving way for more accurate and timely diagnoses.