Facial-Based Autism Spectrum Disorder Diagnosis Using Enhanced Deep Learning Models

Growing interest has been seen in the application of deep learning with facial image analysis to diagnose ASD in recent years. It has been demonstrated that deep learning algorithms can recognize minute facial characteristics linked to ASD. There are, nevertheless, a lot of unmet research needs in this field. In this work, we present a unique deep learning model that uses facial image analysis to diagnose ASD. Modern deep learning architecture Efficient Net serves as the foundation for our model. Additionally, we suggest a novel attention mechanism that enables our model to concentrate on the face traits that are most useful in diagnosing ASD. Using a dataset of 2500 face photos of children with ASD and generally developing kids, we assessed our model. Our model's accuracy of 85% is a considerable improvement above earlier models’ accuracy. Our findings imply that deep learning models may prove to be an effective diagnostic tool for ASD. Our model shows great promise since it can recognize tiny face traits linked to autism spectrum disorders.