Abstract
Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by communication difficulties and repetitive behaviors. Early diagnosis and intervention are critical to improving outcomes for individuals with autism spectrum disorder. In this context, machine learning techniques, especially deep neural networks (DNN), offer effective solutions for pattern prediction and analysis of ASD. This study presents a fivelayer DNN algorithm for behavioral and treatment-based prediction in ASD. Our model uses the power of deep learning to identify complex patterns from multiple sources, including demographic data, medical history, and behavioral assessments collected from multiple individuals with autism spectrum disorder, both positive and negative. The model consists of an input layer, followed by three hidden layers connected to neurons activated by the power function (RELU), and finally oscillation using the sigmoid activation function. ASD process. The model is trained using the Adam optimizer and binary cross-entropy loss function, with additional measures such as early stopping and hyperparameter tuning to prevent overfitting and increase the accuracy of predictions. Our results demonstrate the effectiveness of the proposed DNN model in accurately diagnosing ASD based on comprehensive data and ultimately determining performance metrics. This enables our models to rapidly detect the disease and ultimately contribute to better disease control. K