Design and Implementation of a CNN-based Framework for Predicting Autism Spectrum Disorder

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder, which im-pairs the communication, behavior and interaction. Early diagnosis is an important aspect of enhancing long-term results, but the standard method of diagnosing is time-consuming and requires the expert appraisal. This poses a problem especially in areas where the specialists are inaccessible. A Convolutional Neural Network (CNN) predicts ASD using the AQ-10 behavioral screening data in this research. Various performance measures such as accuracy, precision, recall and F1-score are used to evaluate the model. It is compared to the most popular ma-chine learning models like Support Vector Machines (SVM) and Random Forest. The findings suggest that CNN model is competitive and has better sensitivity in recognizing ASD-positive cases. This allows it to be a convenient early-stage screening tool. The proposed system is not designed to substitute clinical diagnosis but rather aid in the identification of individuals who might need further diagnosis.