Autism Spectrum Disorder Prediction in Kids Using CNN

ABSTRACT Autism is a severe developmental spectrum disorder that puts constraints on communicating linguistic, cognitive, and social interaction skills. Autism spectrum disorder screening detects potential autistic traits in an individual where the early diagnosis shortens the process and has more accurate results. The methods used to predict Autism by doctors involve physical identification of facial features, questioners, Fine motor skills, MRI scans, etc. In the case of widespread developmental abnormalities, the parents feel inadequate to step out in public, and this traditional diagnosis approach requires more time and money. As a result, it is comparable to employing a timely ASD test that aids medical experts and tells individuals whether or not to follow a formal clinical diagnosis. A diagnostic method that can determine a child's likelihood of developing ASD offers the chance to intervene before full symptoms appear.In comparison to the conventional method of diagnosis, the suggested model employs a convolution neural network classifier that helps predict early autistic symptoms in children based on face features in photos with the least amount of time, money, and accuracy. With the aid of the transfer learning technique, the Convolutional Neural Network classifier can identify autism in facial images. Keywords:MachineLearning,Deep Learning,Keras,Facial Features