Detection of Autism Spectrum Disorder Through Orthogonal Decomposition and Pearson Correlation for Feature Selection

Autism Spectrum Disorder (ASD) or simply Autism is a mental disorder that appears within the first three years of life. It leads to mental health-related disorders such as motor skill development, communication, and cognitive skills development. While a challenge there have been enough cases that lead to exceptional scholastic, artistic, and even non- academic skills. Raising a question of recognizing ASD at early stages. Although Scientists and professionals from various domains have worked on improving the methods and reducing the time it takes to diagnose an ASD patient. The use of machine learning and deep learning techniques can help in achieving this task splendidly. The researches done in this field are a few, and due to this, we have tried to do an exhaustive study of the dataset we are working on. We tried to reduce both the time taken to train as well as applied three different techniques. For the study, eight different machine learning models and a deep learning model were used. The models with Pearson features were able to achieve higher accuracy than those with dimensionality reduction techniques. There was no significant improvement in the training times of the reduced datasets as well.