Development of an Interactive Dashboard for Analyzing Autism Spectrum Disorder (ASD) Data using Machine Learning Techniques

Abstract

Autism Spectrum Disease (ASD) is a lifelong neurodevelopmental disorder that impairs a person's capacity to communicate and connect with others. It has an impact on a person's understanding and social relationships. People with ASD also have a wide range of symptoms, such as difficulty communicating with others, repetitive habits, and an inability to operate well in other aspects of daily life. Autism is a "behavioral condition" that can be diagnosed at any age. Symptoms generally develop in the first two years of life. The majority of people are unaware of the disease and so have no way of knowing whether or not someone is disordered. Rather than helping the patient, this usually results in his or her social isolation. ASD is a condition that begins in childhood and continues through adolescence and maturity. About 25 publications were examined in this study on autism spectrum disorder (ASD) prediction using machine learning or data mining. In this thesis, the approaches or algorithms utilized in those works are explained. Furthermore, the data and findings of those articles are evaluated utilizing various techniques and algorithms. Four publicly available non-clinically ASD datasets are used to evaluate the techniques described in those papers. There are 292 instances and 21 attributes in the ASD screening in children dataset. There are 704 instances and 21 attributes in the ASD screening Adult dataset. There are 104 instances and 21 attributes in the ASD screening in teenagers dataset. With 1054 instances and 19 attributes, Toddler is the fourth dataset. An automated dashboard based on the Toddler dataset was built to analyze it and identify insights. Because the focus was on early ASD prediction, this dataset was chosen.