ABSTRACT
Complex manifestations of autism spectrum disorder (ASD) make early detection challenging. Improvement in outcomes is the only means to detect ASD early. Traditional methods used for diagnosis are quite challenging because the symptoms of ASD are very subtle and highly diverse. Machine learning, therefore, presents a great solution all-around depository of multi-dimensional extensive data encompassing behavioral, genetic and clinical information, laying a foundation to trace and design patterns that might reflect ASD characteristics. Advanced neural networks, especially Multi-Layer Perceptrons are crucial for the representation of complex relationships among features and for understanding sequential data in order to identify symptom progression. thereby enhancing the probability of high accuracy and reliability in ASD prediction models and assisting medical personnel in making timely diagnoses.
